CA2513850A1 - Phrase identification in an information retrieval system - Google Patents

Phrase identification in an information retrieval system Download PDF

Info

Publication number
CA2513850A1
CA2513850A1 CA002513850A CA2513850A CA2513850A1 CA 2513850 A1 CA2513850 A1 CA 2513850A1 CA 002513850 A CA002513850 A CA 002513850A CA 2513850 A CA2513850 A CA 2513850A CA 2513850 A1 CA2513850 A1 CA 2513850A1
Authority
CA
Canada
Prior art keywords
phrase
phrases
good
document
documents
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CA002513850A
Other languages
French (fr)
Other versions
CA2513850C (en
Inventor
Anna L. Patterson
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Google LLC
Original Assignee
Google LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Google LLC filed Critical Google LLC
Publication of CA2513850A1 publication Critical patent/CA2513850A1/en
Application granted granted Critical
Publication of CA2513850C publication Critical patent/CA2513850C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/31Indexing; Data structures therefor; Storage structures
    • G06F16/313Selection or weighting of terms for indexing
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10TECHNICAL SUBJECTS COVERED BY FORMER USPC
    • Y10STECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10S707/00Data processing: database and file management or data structures
    • Y10S707/99931Database or file accessing
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10TECHNICAL SUBJECTS COVERED BY FORMER USPC
    • Y10STECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10S707/00Data processing: database and file management or data structures
    • Y10S707/99931Database or file accessing
    • Y10S707/99932Access augmentation or optimizing
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10TECHNICAL SUBJECTS COVERED BY FORMER USPC
    • Y10STECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10S707/00Data processing: database and file management or data structures
    • Y10S707/99931Database or file accessing
    • Y10S707/99933Query processing, i.e. searching
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10TECHNICAL SUBJECTS COVERED BY FORMER USPC
    • Y10STECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y10S707/00Data processing: database and file management or data structures
    • Y10S707/99931Database or file accessing
    • Y10S707/99937Sorting

Abstract

An information retrieval system uses phrases to index, retrieve, organize and describe documents. Phrases are identified that predict the presence of other phrases in documents. Documents are the indexed according to their included phrases. Related phrases and phrase extensions are also identified. Phrases in a query are identified and used to retrieve and rank documents. Phrases are also used to cluster documents in the search results, create document descriptions, and eliminate duplicate documents from the search results, and from the index.

Description

PHRASE IDENTIFICATION IN AN INFORMATION RETRIEVAL S~IS.fELVI
Inventor: Anna L. Patterson Cross Reference to Related Applications [00(?1] Phrase-Based Indexing in an Information Retrieval System, Application No.10~'xxx,xxx, filed on July 2b, 2004;
Phrase-Based Searching in an Information Retrieval System, Application No.
10/xxx,xxx, filed on July 26, 20t?4;
Phrase=Based Personalizatiow of Searches in an Information Retrieval System, Application No. 20/~ac,~ocx, filed on July 26, 2004;
Automatic Taxonomy Generation in Search Results Using Phrases, Application No.
10/xx~,xxx, filed on July 2b, 2004;
Phrase~Based Generation of Document Descriptions, Application Na.10/xxx,xxx, filed on July 26, 2004; and Phrase=Based Detection of Duplicate Documents in an Information Retrieval System, Application No.10/xxx,xxx, filed on July' 26, 2004, all of which are co-owned, and incorporated by reference herein.
Field of the Invention [Q002~ The present invention relates to an information retrieval system for indexing, searching, and classifying documents in a Iarge scale corpus, such as the Internet.

Background of the Invention [0003] Information retrieval systems, generally called search engines, are now an essential tool for finding information in large scale, diverse, and growing corpuses such as the Inhernet. Generally, search engines create an index that relates documents (or ppages") to the individual words present in each document A document is retrieved in response to a query captaining a number of query terms, typically based on having some number of query terms present in the document. The retrieved documents are then ranked according to other statistical measures, such as frequency of occurrence of the query terrcts, host domain, link analysis, and the like. The retrieved documents are then presented to the user, typically in their ranked order, and without any further grouping or imposed hierarchy. In some cases, a selected portion of a text of a document is presented to provide the user with a glimpse of the document's content.
j0004] Direct "Boolean" matching of query terms has well known limitations, and in particular does not identify documents that do not have the query terms, but have related wards. For example, in a typical Boolean syst~n, a search on "Australian Shepherds" would not return documents about. other herding dogs such as Border Collies that do not have the exact query terms. Rather, such a system is likely to also retrieve and highly rank documents That are about Australia (and have nothing to do with dogs}, and documents about "shepherds" generally.
[0005] The problem here is that conventional systems index documents based on individual terms, than on concepts. Concept are often expressed in phrases, such as "Australian Shepherd," "President of the United States," or "Sundance Film Festival".
At best, some prior systems wi71 index documents with respect to a prede6erniined and very limited set of 'known' phrases, which are typically selected by a human operator.
2 Indexing of phrases is typically avoided because of the perceived computational and memory requirements to identify all possible phrases of say three, four, or five or more words. For example, on the assumption that any five words could constitute a phrase, and a large corpus would have at least 2.00,000 unique terms, there would approximately 3.2 x 10~ possible phrases, clearly more than any existing system could store in memory or otherwise programmatically manipulate. A further problem is that phrases continually enter and leave the lexicon in terms of their usage, much more frequently than new individual woxds axe invented. New phrases are ahNays being generatied, from sources such technology, arts, world events, and law. Other phrases will decline in usage over time.
j00<)6] Same existing information retrieval systems attempt to provide retrieval of concepts by using co-occurrence patterns of individual words. In these systems a search on one word, sash as "President" will also retrieve documents that have other words that frequently appear with "President", such as "White" and " House."
While this approach may produce search results having documents that are conceptually related:at the level of individual words, it does not typically capture topical relatianships that inhere between co-occurring phrases.
(OOU7J Accordingly, there is a need for an information retrieval system and methodology that can comprehensively identify phrases in a Iarge scale corpus, index documents according to phrases, search and rank documents in accordance with their phrases, and provide additional clustexing and descriptive information about the documents.
3 Surilmaiw of the Invention j0008J An information retrieval system and methodology uses phrases to index;
search, raxdc, and describe documents in the document collection. The system is adapted to identify phrases that have sufficiently frequent and/or distinguished usage in the document collection to indicate that they are °valid" or "good"
phrases. In this manner multiple word phrases, for example phrases of four, five, or more terms, can be identified. This avoids the problem of having to identify and loader every possible phrases resulting from the all of the possible sequences of a given number of words.
j0009j The system is further adapted to identify phrases that are related to each other, based on a phrase's ability to predict the presence of other phrases in a document.
More specifically, a prediction measure is used that relates the actual co-occurrence rate of two phrases to an expected co-occurrence rate of the two phrases.
Tnformation gain, as the ratio of actual co-occurrence rate to expected co-occurrence rate, is one such prediction measure. Two phrases are related where the prediction measure exceeds a predetermined threshold In that case, the second phrase has si~ificant information gain with respect to the first phrase. Semantically, related phrases w~l be those that are commonly used to discuss or describe a given topic or concept, such as "President of the United:States" and "White House." For a given phrase, the related phrases can be ordered according to their relevance or signz~icance based on their respective prediction measures.
~0010j An information retrieval system indexes documents in the document collection by the valid or good phrases. For each phrase, a posting list identifies the documents that contain the phrase. In addition, for a given phrase, a second list, vector, or other structure is used to store data indicating which of the related phrases of the
4 given phrase are also present in each document containing the given phra~. In this nlannei, the system can readily identify not only which documents contain which phrases in response to a search query, but which documents also contain phrases that are related to query phrases, and thus more likely to be specifically about the topics or concepi~ expressed in the query phrases.
[001I] The use of phrases and related phrases further provides for the creation and use of clusters of related phrases, which represent semantically meaningful groupings o~ phrases. Qusters are identified from related phrases,that have very high prediction measure between all of the phrases in the cluster. Clusters can be used to organize the results of a search, including selecting which documents to include in the search results and theix oxder, as well as eliminating documents from the search results.
[0012] The information retrieval system is also adapted to use the phrases when searching for document's in response to a query. The query is processed to identify any phrases that are present in the query, so as to retrieve the associated posting lists for the query phrases, and the related phrase information. In addition, in some instances a user may enter an incomplete phrase in a search query, such as "President of the".
Incomplete phrases such as these may be identified and replaced by a phrase ext~eension, such as "President of the United States." This helps ensure that the user's most likely search is in fact executed.
j0013] The relahed phrase infornination may also be used by the system to .
identify or select which documents to include in the search result The related phrase information indicates for a given phrase and a given document, which related phrases of the given phrase are present in the given document. Accordingly, for a query containing two query phrases, the posting list for a first query phrase is processed to identrfy documents containing the first query phrase, and then the related phrase information is processed to identify which of these documents also contain the second query phrase.
These latter documents are then included in the search results. 'This eliminates the need for the system to then separately process the posting list of the second query phrase, thereby providing faster search times. Of course, this approach may be extended to any number of phrases in a query, yielding in significant computational and timing savings.
[0014] The system may be further adapted to use the phrase and related phrase information to rank documents in a set of search results. The related phrase information of a given phrase is preferably stared in a format, such as a bit vector, which expresses the reiative significance of each related phrase to the given phrase. For example, a related phrase bit vector has a bit for each related phrase of the given phrase, and the bits are ordered according to the prediction measures (e.g., information gain) for the,related phrases. The most significant bit of the related phrase bit vector are associated with the related phrase having the highest prediction measure, and the least significant bit is associated with the related phrase having a lowest prediction measure.
In this manner, for a given document and a given phrase, the related phrase information can be used to score the document. The value of the bit vector itself (as a value} may be used as the document score. In this manner documents that contain high order related phrases of a query phrase are more likely to be topically related to the query than those that have low ordered related phrases. The bit vector value may also be used as a component in a more complex scoring function, and additionally may be weighted. The documents can then be ranked according to their document scores.
[0015] Phrase information may also be used in an information retrieval system to personalize searches for a user. A user is modeled as a collection of phrases, for example, derived from documents that the user has accessed {e.g., viewed on screen, printed, stored, etc.). More particularly, given a document accessed by user, the related phrases that are present in this document, are included in a user model or profile.
During subsequent searches, the phrases in the user model are ~ed~ to filter the phrases of the search query and to weight the document scores of the retrieved docamen~.
[OD1GJ Phrase information may also be used in an information retrieval system to create a description of a document, for example the documents included in a set of search iesults. Given a search query, the system identifies the phrases present in the query, along with their related phrases, and their phrase extensions. For a given document, each sentence of the document has a count of how many of the query phrases, related phrases, and phrase extensions are present in the sentence.
The sentences of document can be ranked by these counts {individually or in combination), axed some number of the top ranking sentences (e.g., five sentences) are selected to form the document description. The document description can then be presented to the user when the document is included in search results, so that the user obtains a better understanding of the document, relative to the query.
j00~.7] A further refinement of this process of generating document descriptions allows the system to provide personalized descriptions, that reflect the interests of the user. As before, a user model stores information identifying related phrases that are of interest to the user. This user modek is intersected with a list of phrases related to the query phrases, to identify phrases connnnon to both groups. The common set is then ordered according to the rekated phrase information. The resulting set of related phrases is then:used to rank the sentences of a document according to the number of instances of these related phrases present in each document A number of sentences having the highest number of common related phrases is selected as the personalized document description.
[0018) An information retrieval system may also use the phrase information to identify and eliminate duplicate documents, either while indexing {crawling) the document collection, or when processing a search query. For a given document, each sentence of the document has a count of how many related phrases are present in the sentence. The sentences of document can be ranked by this count, and a number of the top ranking sentences {e.g., five sentences) are selecfied to form a document description.
This description is then stored in association with the document, for example as a string or a hash of the. sentences. >3urix~g index~g, a newly crawled document is processed in the same manner to generate the document description, The new document description can be matched (e.g., hashed) against previous document descriptions, and if a match is found, then the new document is a duplicate. Similarly, during preparation of the results of a search query, the documents in the search result set can be processed to eliminate duplicates.
[oolgl The present invention has further embodiments in system and software architectures, computer program products and computer implemented methods, and computer generated user interfaces and presentations.
[0020] The foregoing are just some of the features of an information retrieval system:and methodology based on phrases. Those of skill in the art of informatlon retrieval will appreciate the flexibility of generality of the phrase information allows for a large variety of uses and applications in indexing, document annotation, searching, ranking, and other areas of document analysis and processing.

Brief Descn~tion of the Drawings [002IJ FIG.1 is block diagram of the softwaxe architecture of one embodiment of the present invention.
[0022] FIG. 2 illustrates a method of identifying phrases in documents.
(0023] FIG. 3 illusixafies a document with a phrase window and a sernndary window.
[0024] FIG. 4 illustrates a method of identifying related phrases.
[0025] FIG. 5 illustrates a method of indexing documents for related phrases.
[0026] FIG. 6 illustrates a method of retrievnng documents l~sed on phrases.
[0027] FIG. 7 illustrates operations of the presentation system to present search results.
[0028] FIGS. 8a and 8b illustrate relationships between referencing and referenced documents.
[0029] The figures depict a pzeferred embodiment of the present invention far purposes of illustration only. One skr~led in the art will readily recognize from the following discussion that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the invezitian described herein.
Detailed Description of the Invention I. S~~stem.Overview [0030] Referringnow to FIG.1, there is shown the software archifiecture of an embodiment of a search systexn 100 in accordance with one embodiment of present invention. In this embodiment, the system includes a indexing system 110, a search system '120, a presentation system 130, and a front end server 140.
j~3lJ T'he indexing system 110 is responsible for identifying phrases in documents, and indexing documents according to their phrases, by accessing various websites 190 and other document collections. The front end server 140 receives queries from a user of a client 170, and provides those queries fio the search system 120. The search system 220 is responsible for searching for documents relevant to the search query (search results), including iden#ifying any phrases in the search query, and then ranking the documents in the search results using the presence of phrases to influence the ranking order. The search system 120 provides the search results to the presentation systemM30. The presentation system 130 is responsible for modifying the search results including removing near duplicate documents, and generating topical descriptions of documents, and providing the modified search results back fio the front end server 140, which provides the results to the client 170. The system 100 further includes an index 1~0 that stores the indexing information pertaining to documents, and a phrase data store lfi0 that stores phrases, and related statistical information.
0032] In the context of this application, "documents" are understood ho be any type of :media that can be indexed and retrieved by a search engine, including web documents, images, multimedia files, text documents, PDFs or other image formatted files, and so forth. A document may have one or more pages, partitions, segments or other components, as appropriate to its content and type. Equivalently a document may be referred to as a "page," as commonly used to refer to documents on the Internet. No limitation as to the scope of the invention is implied by the use of the generic term "documents." The search system 100 operates over a large corpus of documents, such as the Internet and World Wide Web, but can likewise be used in more limited collections, such as~for the document collections of a library or private enterprises. In either context; it will be appreciated that the documents are typically distr~ibubed across many different computer systems and sites. Without loss of generality then, the docttment~s generally, regardless of format ox location (e.g., which website or database) will be collectively referred to as a corpus or document collection. Each document has an associated identifier that uniquely identifies the document; the identifier is preferably a URL, but other types of identifiers (e.g., document numbers) may be used as welt. In this disclosure, the use of URLs to identify documents is assumed.
XI. Indexing S~ystem [0033] In one embodiment, the indexing system 210 provides three prinlaxy functional operations:1) identification of phrases and related phrases, 2) indexing of documents with respect to phrases, and 3) generation and maintenance of a phrase-based taxonomy. Those of skill in the art will appreciate that the indexing system x10 wdl perform other functions as well in support of conveniio~nal indexing functions, and thus these other operations are not farther described herein. The indexing system 110 operates on an index 150 and data repository 2b0 of phrase data These data repositories are further died below.
1. Phrase Identification [0034] The phrase identification operation of the indexing system 110 identifies "good" and "bad" phrases in the document collection that are useful to indexing and searching documents. In one aspect, good phrases are phrases that tend to occur in more than certain percentage of documents in the document collection, and/or are indicated as having a distinguished appearance in such documents, such as delimited by markup tags or other morphological, format, or grammatical markers. Another aspect of good phrases is that they are predictive of other good phrases, and aze not merely sequences of words that appear in the lexicon. For example, the phrase "President of the Unified States" is a phrase that predicts other phrases such as "George Bush"
and "Bill Clinton." However, other phrases are not predictive, such as "fell down the stairs" or "top of the morning," "out of the blue, ' since idioms and colloquisms like these tend to appear with many other different and unrelated phrases. Thus, the phrase identification phase determines which phrases are good phrases and which are bad (i.e., Iackmg in predictive power).
[003] Referring to now FIG. 2, the phrase identification process has the faltowing fun~onal stages:
[003 200: Collect possible and good phrases, along with frequency and co-occurrence statistics of the phrases.
j0037J Za2: Qassify possible phrases to either good or bad phrases based on frequency statistics.
[0038] 204: Prune good phrase list based on a predictive measure derived ~xo:an the co-accurrence statistics.
[0039r Each of these stages will now be described nn further detail.
[0040] The first stage 200 is a process by which the indexing system 120 crawls a set of documents in the document collection, making repeated partitions of the document collection over time. Gne partition is processed per pass. The number of documents crawled per pass can vary, and is preferably about 1,000,000 per par#ition. It is preferred that only previously uncrawled documents are processed in each partition, until all documents have been processed, or some other temination criteria is met. In practice, the crawling continues as new documents are being continually added to the document collection. The following steps are taken by the indexing system 110 for each document that is crawled.
10041] Traverse the words of the document with a phrase window length of r~, where n is a desired maximum phrase length. The length of the window will typically be at least 2, and preferably 4 or 5 terms {words). Preferably phrases include all words in the phrase window, including what would otherwise be characterized as stop words, such as "a", "the," and so forth. A phrase window may be terminated by an end of Iine, a paragraph return, a markup tag, or other indicia of a change in content or format.
[0042] FIG. 3 illustrates a portion of a document 300 during a traversaX, showing the phrase window 302 starting at the word "stock" and extending 5 words to the right.
The first word in the window 302 is candidate phrase i, and the each of the sequences i+I, i+2, i+3, i+4, and i+5 is likewise a candidate phrase. Thus, in this example, the candidate phrases are: "stock", "stock dogs", "stock dogs for", "stock dogs for the", "stack dogs for the Basque", and "stock dogs for the Basque shepherds".
(0043] In each phrase window 302, each candidate phrase is checked in turn to determine if it is already present in the good phrase list 208 or the possible phrase list 20&. If the candidate phrase is not present in either the good phrase list 208 or the possible phrase list 206, then the candidate has already been determined to be "bad" and is skipped.
j0044] If the candidate phrase is in the good phrase list 208, as entry g~, then the.
index 150 entryfor phrase g~ is updated to include the document (e.g., its URL
or other document identifier), to indicate that this candidate phrase g; appears in the current document. An entry in the index 150 for a phrase ~ (or a term) is referred to as the posting'Iist of the phrase g;. The posting list includes a list of documents d {by their document identifiers, e.g. a document number, or alternatively a URL) in which the phrase occurs.
[0045] In addition, the co-occurrence matrix 212 is updated, as further explained below. :1n the very first pass, the good and bad lists will be empty, and thus, most phrases will tend to be added to the possible phrase list 206.
[0046] If the candidate phrase is not in the good phrase list 248 then it is added to the possible phrase list 24G, unless it is already present therein. Each entry p on the possible phrase list 206 has three associated counts:
[0047j P(p): Number of documents on which the possible phrase appears;
[0048] S(p): Number of all instances of the possible phrase; and [0049] M(p): Number of interesting instances of the possible phrase. An instance of a possible phrase is "interesting" where the possible phrase is distinguished from neighboring content in the document by grammatical or format markers, for example by being iri boldface, or underline, or as anchor text in a hyperlink, or in quotation marks.
These (and other) distinguishing appearances are indicated by various I-1TML
markup language tags and grammatical markers. These statistics are maintained for a phrase when it is placed on the good phrase list 208.
[0050] In addition the various lists, a co-occurrence matrix 212 {G) for the good phrases is maintained. The matrix G has a dimension of m x m, where m is the number of good phrases. Each entry G(j, k) in the matrix represents a pair of good phrases (g~, gx). The co-occurrence matrix 212 logically (though not necessarily physically) maintains three separate counts for each pair (g;, gk) of good phrases with respect to a secondary window 304 that is centered at the current word i, and extends ~/- h words. In one embodiment, such as illustrated in FIG. 3, the secondary window 304 is 30 words. The co-occurrence matrix 212 thus maintains:
[0051] R(j,k): Raw Co-occurrence count. The number of times that phrase g~
appears in a secondary window 304 with phrase gt;
[0(152] D(jk): Disjunctive Interesting count. The number of times that either phrase g~ or phrase gk appears as distinguished text in a secondary window;
and [0053) C(j,k): Conjunctive interesting count the number of times that both g~
and phrase ~gx appear as distinguished text in a secondary window. The use of the cove interesting count is particularly beneficial to avoid the circumstance where a phrase (e.g., a copyright notice) appears frequently in sidebars, footers, or headers, and thus is not actually predictive of other text.
[0054] Referring to the example of FIG. 3, assume that the "stock dogs" is on the good phrase list 208, as well as the phrases "Australian Shepherd" and "Australian Shepard Club of America". Both of these latter phrases appear within the secondary window 304 around the current phrase "stock dogs". However, the phrase "Australian Shepheid Club of America" appears as anchor text for a hyperlink (indicated by the underline) to website. Thus the raw co-occurrence count for the pair {"stock dogs", "Australian Shepherd"} is incremented, and the raw occurrence count and the disjunctive interesting count for {"stock dogs", "Australian Shepherd Club of AmericaN}
are both incremented because the latter appears as distinguished text.
(0055] The process of traversing each document with both the sequence window 302 and the secondary window 304, is repeated for each document in the partition.
[00561 Once the documents in the partition have been traversed, the next stage of the indexing operation is to update 202 the good phrase list 208 from the poss~le phrase list 206. A possible phrase p on the possible phrase list 20G is moved to the good phrase Zist 208 if the frequency of appearance of the phrase and the number of documents that the phrase appears in indicates that it has sufficient usage as semantically meaxtixigfnl phrase.
[0057] In one embodiment, this is tested as follows. A possible phrase p is removed from the possible phrase list 20b and placed on the good phrase list 208 if:
[0058j a} P(p) > 10 and S(p) > 20 {the number of documents containing phrase p is more than 10, and the number of occurrences of phrase p is more then 20};
or (OQ59~ b} M(p) > 5 tthe number of interesting instances of phrase p is more than [0Q60] These thresholds am scaled by the number of documents in the partition;
for example if 2,000,000 documents are crawled in a partition, then the thresholds are approximately doubled. Of course, those of skill in the art will appreciate that the specific values of the thresholds, or fihe logic of testing them, can be varied as desired.
[0061] If a phrase p does not qualify for the good phrase list 2D8, then it is checked for qualification for being a bad phrase. A phrase p is a bad phrase if:
[0062] a) number of documents containing phrase, P{p) < 2; and j0063] b) number of interesting instances of phrase, M{p) = 0.
10064] These conditions indicate that the phrase is both infrequent, and not used as indicative of significant content and again these thresholds may be scaled per number of documents in the partition.
j00651 It should be noted that the good phrase list 208 will naturally include individual words as phrases, in addition to mufti-word phrases, as described above.

This is because each the first word in the phrase window 302 is always a candidate phrase,:and the appropriate instance counts will be accumulated. Thus, the indexing system X10 can automatically index both individual words (i.e., phrases with a single word) and multiple word phrases. The good phrase list 208 will also be considerably shorter than the theoretical maximum based an all possible combinations of m phrases.
In typical embodiment, the good phrase list 208 will include about 6.5x105 phrases. A
list of bad phrases is not necessary to store, as the system need only keep track of possible and good phrases.
(0066j By the final pass through the document collection, the list of possible phrases will be relatively short, due to the expected distribution of the use of phrases in a large corpus. Thus, if say by the 10~ pass (e.g.,10,000,000 documents}, a phrase appears for the very first time, it is very unlikely to be a good phrase at that time. It may be new phrase just coming into usage, and thus during subsequent crawls becomes increasingly common. In that case, its respective counts will increases and may ultimately satisfy tire thresholds for being a good phrase.
I0067j The third stage of the indexing operation is to prune 204 the good phrase list 208 using a predictive measure derived from the co-occurxence matrix 212.
Without pruning; the good phrase list 208 is lr'kely to include many phrases that while legitimately appearing in the lexicon, themselves do not sufficiently predict the presence of other phrases, or themselves are subsequences of longer phrases. Removing these weak good phrases results in a very robust likely of good phrases. To identify good phrases, a predictive measure is used which expresses the increased likelihood of one phrase appearing in a document given the presence of another phrase. This is done, in one embodiment, as follows:

[0058] As noted above, the co-occurrence matrix n2 is an m x m matrix of storing data associated with the good phrases. Each row j in the matrix represents a good phrase g~ and each column k represented a good phrase gx. For each good phrase gJ, an expected value E(g;) is computed. The expected value E is the percentage of documents in the collection expected to contain g;. 'This is computed, for example, as the ratio of the number of documents containing g; to the total number T of documents in the collection that have been crawled: P(j)/T.
[0069) As rioted above, the number of documents containing g; is updated each time g; appears in a document. The value for E(g;) can be updated each time the counts for g~ are incremented, or during this third stage.
[0070) Next, for each other good phrase gk {e.g., the columns of the matrix), it is determined whether gi predicts gk. A predictive measure for g~ is determined as follows:
[00TTJ t) compute the expected value E{gk). The expected co-occurrence rate E(jk) of g; and gx, if they were unrelated phrases is then E(g~)*E(gx);
[0072) ii) compute the actual co-occurrence rate A(j,k) of gT and gt. This is the raw co~ccurrence count R(j, k) divided by T, the total number of documents;
[0073) iii) ~ is said to predict gx where the actual co-occuzrence rate A(jk) exceeds the expected co-occurrence rate E{j,k) by a threshold amount.
[0074) In one embodiment, the predictive measure is information gain. Thus, a phrase g~ predicts another phrase gk when the information gain 1 of g~ in the presence of g~ exceeds a threshold. In one embodiment, this is computed as follows:
ill Z~k) = Atl~k)/) j0076J And goad phrase gJ predicts good phrase gx where:

j(?OT7] I(j,k) > Information Gain threshold.
j0078] In one embodiment, the information gain threshold is 1.5, but is prefera~Iy between T.1 and 1.7. Raising the threshold over L0 serves to reduce the possibility that two otherwise unrelated phrases co-occur more than randomly predicted.
[4079] As noted the computation of information gain is repeated for each column'k of the matrix G with respect to a given row p. Once a row is complete, if the information gain for none of the good phrases gx exceeds the information gain thzes~hold, then this means that phrase g~ does not predict any other good phrase. In that case' g~ is removed from the good phrase list 208, essentially becoming a bad phrase. Note that the coluW n j for the phrase ~ is not removed, as this phrase itself may be predicted by other good phrases.
This step is concluded when all rows of the co-occurrence matrix 212 have been evaluated.
[00$1] The final step of this stage is to prune the good phrase list 208 to remove incomplete phrases. An incomplete phrase is a phrase that only predicts its phrase extensions, and which starts at the Ieft most side of the phrase ti.e., the beginning of the phrase): The "phrase exfiension" of phrase p is a super-sequence that begins with phrase p. For eXample, the phrase "President of" predicts "President of the iJnited States", "President of Mexico", "President of AT&T", etc. All of these latter phrases are phrase extensions of the phrase "President of" since they begin with "President oY' and are super-sequences thereof.
[0082] Accordingly, each phrase g~ remaixiing on the good phrase list 208 will predict sonne number of other phrases, based on the information gain threshold previously discussed. Now, for each phrase g; the indexing system 11f) performs a string match with each of the phrases g~ that is predicts. The string match tests whether each predicted phrase gx is a phrase extension of the phrase gl. if all of the predicted phrases gx are phrase extensions of phrase g~, then phrase gj is incomplete, and is removed from the good phrase Iist 208, and added to an incomplete phrase list 216.
Thus, if there is at least one phrase gk that is not an extension of g~, then g,; is complete, and maintained in the good phrase list 208. For example then, "President of the United" is an incomplete phrase because the only other phrase that it predicts is °President of the United States" which is an extension of the phrase.
jD083] The incomplete phrase list 216 itself is very useful during actual searching. When a search query is received, it can be compared against the incomplete phase list 216. If the query (or a portion thexeof) matches an entry in the list, then the search system '220 can lookup the most likely phrase extensions of the incomplete phrase {the phrase extension having the highest information gain given the incomplei~e phrase), and suggest this phrase extension to the user, or automatically search on the phrase extension. For example, if the search query is "President of the United," the search system -12U can automatically suggest to the user "President of the United States" as the search query.
j0084] After the last stage of the indexing process is completed, the good phrase list 208 will contain a large. number of good phrases that have been discovered in the corpus. Each of these good phrases will predict at least one other phrase that is not a phxase extension of it. That is, each good phrase is used with sufficient frequency and independence to represent meaningful concepts or ideas expressed in the corpus. Unvlce existing systems which use predetermined or hand selected phrases, the good phrase list zo reflects phrases that actual are being used in the corpus. Further, since the above process of crawling and indexing is repeated periodically as new documents are added to the document collection, the indexing system 210 automatically detects new phrases as they enter the lexicon.
2. Identification of Related Phrases and Qusters of Related Phrases [0085j Referring to FIG. 4, the related phrase identification process includes the following functional operations.
[00$6] 400: Identify related phrases having a high information gain value.
[00$7] 402: Identify clusters of related phrases.
[40$$] 404: Store cluster bit vector and cluster nc~m't~r.
[00$9j Each of these operations is now descn'bed in detail.
[0090] First, recall that the co-occurrence matrix 222 contains good phrases g~, each of xvhich predicts at Ieast one other good phrase gk with an information gain greater than the'information gain threshold. To identify 400 related phrases then, for. each pair of good phrases {g~, gx) the information gain is compared with a Related Phrase threshold, e.g., 200. That is, g~ and gx are related phrases where:
[~91j I(gf, gr) > 140.
[0092] This high threshold is used to identify the co-occurrences of good phrases that are well beyond the statistically expected rates. Statistically, it means that phrases g;
and gk co-occur 100 times more than the expected co-occurrence rate. For example, given the phrase "Monica Lewinsky" in a document, the phrase "Bill Qinton" is a 1~
times more likely to appear in the same document, then the phrase "Bill Clinton" is likely to appear an any randomly selected document. Another way of saying this is that the accuiacy of the predication is 99.999 because the occurrence rate is 100:1.

[0093] Accardingly, any entry (g;, gx} that is less the Related Phrase threshold is zeroed out, indicating that the phrases g;, gk are not related. Any remaining entries in the co-occurrence matrix 212 now indicate all related phrases.
[0D94] The columns g~ in each row g~ of the co-occurrence matrix 212 are then sorbed by the information gain values I(ga, gx), so that the related phrase g~
with the highest information gain is listed first. This sorting thus identifies for a given phrase g;, which other phrases are most likely related in terms of information gain.
~U095) The next step is to determine 402 which related phrases together form a cluster of related phrases. A cluster is a set of related phrases in which each phrase has high information gain with respect to at least one other phrase. In ane embodiment, clusters are identified as follows.
[0096] In each row gr of the matrix, there will be one or more other phrases that are related to phrase g;. This set is related phrase set Rj, where R={g~, gi, ... g",}.
(0097) For each related phrase m in R~, the indexing system 210 detemnines if each of the other related phrases in R is also related to g;. Thus, if I(g~, gi) is also non-zero, then gi, gk, and gi are part of a cluster. This cluster test is repeated for each pair (g~, ga,} in R.
[0098] For example, assume the good phrase "Bill Qint~on" is related to the phrases "President', "Mvnica Lewinsky", because the information gain of each of these phrases with respect to "Bill Clinton" exceeds the Related Phrase threshold.
Further assume chat the phrase "Monica Lewinsky" is related to the phrase "purse designer".
These phrases then form the set R. To determine the clusters, the indexing system 120 evaluates the information gain of each of these phrases to the others by determining their corresponding information gains. Thus, the indexing system 110 determines the information gain 1("President", "Monica Lewinsky'), I("President", "purse designer"), and so #orth, for all pairs in R. In this example, "Bill Clinaon,' "President", and "Monica Lewinsky" form a one cluster, "Bill Ginton," and "President" form a second cluster, and "Monica Lewinsky" and "purse designer" form a third cluster, and "Monica Lewinsky", "Bill Clinton," and "purse designer" form a fourth cluster. This is because while "Bill Qintori" does not predict "purse designer" with sufficient information gain, "Monica Le~vinsky" does predict both of these phrases.
(0099] To record 404 the cluster information, each cluster is assigned a unique cluster number (cluster ID). This information is then recorded in conjunction with each good phrase g~.
(OIOOJ In one embodiment, the cluster number is determined by a cluster bit vector that also indicates the orthogonality relationships between the phrases. The cluster bit vector is a sequence of bits of length n, thewumber of good phrases in the good phrase list 208. For a given good phrase g;, the bit positions correspond to the sorbed related phrases R of g;. A bit is set if the related phrase gk in R is in the same cluster as phrase g;. More generally, this means that the corresponding bit in the cluster bit vectox is set if thexe is infoxmation gain in either direction between g;
and gk.
(010I] The duster number then is the value of the bit string fihat results.
This implementation has the property that related phrases that have multiple ox one-way information gain appear in the same cluster.
(~i02J An example of the cluster bit vectors are as follows, using the above phrases:
Monica purse Quster Bill Clinton President Lewinsky designer l ID

Bill Qi~ton1 1 2 0 14 President 1 1 0 0 22 Monica ~w~ 1 0 2 1 11 purse d~I ~ 0 0 1 1 3 [01.Q3] To summarize then, after this process there will be iderdified for each good phrase g;, a set of related phrases R, which are sorted in order of information gain 1{g~, gk} from highest to lowest. In additior~, for each good phrase g~, there will be a cluster bit vertor, the value of which is a cluster number identifying the primary dustier of which the phrase g~ is a member, and the orthogonality values {1 or 0 for each bit position} indicating which of the related phrases in R are in connmon clusters with g~.
Thus in the above example, "Bill Qinton", "President", and "Monica Lewinsky"
are in cluster 14 based on the values of the bits in the row for phrase "Bill Clinton~.
[004] To store this information, two basic representations are available.
First, as indicated above, the information may be stored in the co-occurrence matrix 222, wherein:
j0105~ entry Gjrow j, col. k} _ (1{j,k}, clusterNumber, dusterBitVector}
j0106) AltemativeIy, the matrix representation can be avoided, and all information stored in the good phrase list ZOB, wherein each row therein represents a good phrase gr:
j0107J Phrase rowl = list [phrase gx, (i{j,k}, clusterNumber, dusterBitVector)j.
j010$j This appxoach pxovides a useful organization for clusters. First, rather than a strictly--and often arbitrarily-defined hierarchy of topics and concepts, this approach recognizes that topics, as indicated by related phrases, form a complex graph of rela 'tuynslups, where some phrases are related to many other phrases, and some phrases:have a more limited scope, and where the relationships can be mutual (each phrase predicts the other phrase) or one-directional (one phrase predicts the other, but not vice versa). The result is that clusters can be characterized "local" to each good phrase, and some clusters will then overlap by having one or more common related phrases:
[0I09] For a given good phrase g~ then the ordering of the related phrases by information gain provides a taxonomy for naming the clusters of the phrase:
the cluster name is the name of the related phrase in the cluster having the highest information [dlld] The above process provides a very robust way of identifying significant phrases that appear in the document collection, and beneficially, the way these relafied phrases. are used together in natural "clusters" in actual practice. As a result, this data-driven clustering of related phrases avoids the biases that are inherent in any manually directed "ediix~rial" selection of related terms and concepts, as is common in many systems.
3. Indexing Documents with Phrases and Related Phrases [dill] Given the good phrase list 208, including the information pertaining to related phrases and clusters, the next functional operation of the indexing system 210 is to index documents in the document collection with respect to the good phrases and dusters, and store the updated information in the index 250. FIG. 5 illustrates this process; in which there are the following functional stages for indexing a document [dlT2j 500: Post document m the posting lists of good phrases found in the document.

[0113] 502: Update instance counts and related phrase bit vector for related phases end secondary related phrases.
[0I14] 504: Annotate documents with related phrase information.
[Q115] 506: Reorder index entries according to posting Iist size.
[01I6] These stages are now described in further detail.
[OIZ7j A set of documents is traversed or crawled, as before; this may be the same or.a different set of documents. For a given document d, traverse 500 the document word by word with a sequence window 302 of length t:, from position i, in the manner:descn'bed above.
j011$] In a given phrase window 302, identify all good phrases in the window, starting at position i. Each good phrase is denoted as g;. Thus, g1 is the first good phrase, g2 would be the second good phrase, and so forth.
[Q119] For each good phrase g; (example g2 "President" and g4 "President of ATT") post the document identifier (e.g., the URL) to the posting list for the good phrase g; in the index 150. This update identifies that the good phrase g; appears in this specific document.
j0I20] In one eufbodiment, the posting list for a phrase g~ takes the following logical form:
jd121] Phrase gJ: list: (document d, just: related phase counts] [related phrase information]) [(f122] For each phrase g~ there is a list of the documents d on which the phrase appears. For each document, there is a list of counts of the number of occurrences of the related phrases R of phrase gl that also appear in document d.
2b (0123 In one embodiment, the related phrase information is a related phase bit vector. 'This bit vector may be characterized as a obi bit"' vector, in that for each related phrase gx there are two bit positions, gk-2, grr2. The first bit position stores a flag indicating whether the related phrase g,~ is present in the document d (i.e., the count for gk in doruinent d is greater than 0). The second bit position stores a flag that indicates whether a related phrase gi of gk is also present in document d. The related phrases gt of a related phrase gx of a phrase g~ are herein called the "secondary related phrases of gi "
The rnunts and bit positions correspond to the canonical order of the phrases in R
(sorted in order of decreasing information gain). This sort order has the effect of making the relabed phrase gr~ that is most highly predicted by g; associated with the most significant bit of the related phrase bit vector, and the related phrase gi that is least predicted by g; associated with the least significant bit.
(U1,24~ It is useful to note that for a given phrase g, the length of the related phrase bit vector, and the association of the related phrases to the individual bits of the vectox, Twill be the same with respect to all documents containing g. This implementation has the properly of allowing the system to readily compare the related phrase bit vectors for any (or all) documents rnntaining g, to see which documents have a given related phrase. This is beneficial for facilitating the search process to identify documents in response to a search query. Accordingly, a given document will appear in the posting lists of many different phrases, and in each such posting list, the related phrase vector for that document will be specific to the phrase that owns the posting list.
This aspect preserves the locality of the related phrase bit vectors with respect to individual phrases and documents.

[0125] Accordingly, the next stage 502 includes traversing the secondary window. 30~ of the current index position in the document (as before a secondary window. of +/- K terms, for example, 30 terms), for example from i-K to i+K.
For each related lihrase g~ of g; that appears in the secondary window 304, the indexing system 1~0 increments the count of gt with respect to document d in the related phrase o~ If g; appeals later in the document, and the related phrase is found again within the later secondary window, again the count is incremented.
[0126] As noted, the corresponding first bit g~-1 in the related phrase bit map is set based on the count, with the bit set to 1 if the count for gk is >D, or set to D if the count equals 0.
[0127f Next, the second bit, g~r2 is set by looking up related phrase gk in the index 150, identifying in gk's posting list the entry for document d, and then checking the secondary related phrase counts (or bits) for gk for any its related phrases.
If any of these secondary related phrases caunts/bits are set, then this indicates that the secondary related phrases of g~ are also present in document d.
[0I2$] Why document d has been complehely processed in this manner, the indexing system 1~0 will have identified the following:
[0129] l) each good phrase g~ in document d;
10130] ii) for each good phrase g~ which of its related phrases g~ are present in document d;
[0131] iii) for each xelated phrase g,Y present in document d, which of its related phrases g, (the secondary related phrases of g;) are also present in document d a) Determinin~the Topics for a Document j0132] The indexing of documents by phrases and use of the clustering information provides yet another advantage of the indexing system 110, which is the ability to determine the topics that a document is about based on the related phrase information.
[0133] Assume that for a given good phrase ga and a given document d, the posting list entry is as follows:
j0134] g: document d: related phrase counts= {3,4,3,0,0,2,7,1,0}
10135] related phrase bit vector:={ 311110 00 00101010 01}
[0136] where, the related phrase bit vector is shown in the bi-bit pairs.
[0137) From the related phrase bit vector, we can determine primary and secondary topics for the document d. A primary topic is indicated by a bit pair (1,1), and a secondary topic is indicated by a bit pair (1,0). A related phrase bit pair of (1,1) indicates that both the related phrase gk for the bit pair is present in document d, along the secondary related phrases gi as well. This may be interpreted to mean that the author of the document d used several related phrases gl, gk, and gr together in drafting the document. A bit pair of (1,0} indicates that both g~ and gk are present, but no further secondary related phrases from gx are present, and thus this is a Iess significant topic.
b) Document Annotation for Improved Ranking j0138] A further aspect of the indexing system 110 is the ability to annotate each document d during the indexing process with information that provides for improved ranking during subsequent searches. The annotation process 50~ is as follows.

[U139] A given document d in the document collection may have some number of outl'iriks to other documents. Bach outlink (a hyperlink) includes anchor text and the document identifier of the target document. For purposes of explanation, a current document d being processed will be referred to as URLO, and the target document of an outlink an document d will be referred to as URL1. For Iater use in ranking documents in search results, for every link in LtRLO, which points to some other URLi, the indexing system 110 creates an outlink score for the anchor phrase of that link with respect to URLO, apd. an inIink score for that anchor phrase with respect to URLi. That is, each Iink in the document collection has a pair of scores, an outlink score and an inlink score.
These scores are computed as follows.
[0140] On a given document URLO, the indexing system 110 identifies each outlink to another document URL1, in which the anchor text A is a phrase in the good phrase list 208. FIG. 8a illustrates schematically this relationship, in which anchor text ~.A.. m d~~t ~~ ~ used in a hyperlink 8~.
[0141) In the posting list far phrase A, URLO is posted as an outlink of phrase A, and URL1 is posted as an inlink of phrase A. Far URLO, the related phrase bit vector is completed as described above, to identify the related phrases and secondary related phrases of A pxesent in URLO. This related phrase bit vector is used as the outiink score for the link from URLO to U12L2 containing anchor phrase A.
[0142] Next, the inlink yore is determined as follow. For each inlink to LTRLI
containing the anchor phrase A, the indexing system 110 scans URLI, and deternvines whether. phrase A appears in the body of URLI. If phrase A not only points to URLI
(via a outIink on URLO~, but also appears in the content of URLI itself, this suggests that URL1 can be said to be intentionally related to the concept represented by phrase A.

FIG. 8b Eltustrates this case, where phrase A appears in both URLO (as anchor text) and in the bbdy of URL1. In this case, the related phrase bit vector for phrase A
for UR.L1 is used as the inlink score for the link from URLO to URL1 containing phrase A.
(0143) If the anchor phrase A does not appear in the body of URL1 (as in FIG.
8a), thezt a different step is taken to determine the inlink score. In this case, the indexing system 110 creates a related phrase bit vector far URL1 for phrase A
tas if phrase A was present in URL1) and indicating which of the related phrases of phrase A
appear nn U'RL1. This related phrase bit vector is then used as the inlink score for the link from URLO to URL1.
j0144) For example, assume the following phrases are initially present in L'IRL,O
and URLl:
Anchor Phrase Related Phrase Bit Vector Australian blue red agility Document Sh herd Aussiemerle merle tricolortraixun URL1 1 0 1--~ - 1 ~7-~ - _ (0145) (In the above, and following tables, the secondary related phrase bits are not shown). The URLO mw is the outlink score of the link from anchor text A, and the URL1 row is the inlink score of the link Here, URLO contains the anchor phrase "Australian Shepard" which targets URL1. Of the five related.phrases of "Australian Shepard", only one, "Aussie" appears in URLO. Intuitively then, UItLO is only weakly about Australian Shepherds. URL1, by comparison, not only has the phrase "Australian Shepherd" present in the body of the document, but also has many of the related phrases present as well, "blue merle,° "red merle," and "tricolor."
Accordingly, because the anchor phrase "Australian Shepard" appears in both URLO and URL1, the outlink score for URLO, and the inlinlc score for URL1 are the respective rows shown above.
[0146] The second case described above is where anchor phrase A does not appear ~ URL1. In that, the indexing system 110 scans URL1 and determines which of the related phrases "Aussie," "blue merle," "red merle,' "tricolor," and "agility training" are present in URL1, and creates an related phrase bit vector accordingly, for example:
Anchor Phrase Related Phrase Bit Vector Australian blue red agility Document She herd Aussiemerle merle tricolortr uRLO 1 1 0 0 0 0 ~URL1 - ~ ~ ~ ~ ~ 1 1 0 - I r [0147] Here, tins shows that the URL1 does not contain the anchor phrase "Australian Shepard", but does contain the related phrases "blue merle", "red merle", and "tricolor".
[4148] This approach has the benefit of entirely preventing certain types of manipulations of web pages (a class of documents) in order to skew the results of a search. Search engines that use a ranking algorithm that relies on the number of links that point to a given document in order to rank that document can be "bombed"
by artificially creating a large number of pages with a given anchor text which then point to a desired page. As a result, when a search query using the anchor text is entered, the desired page is typically returned, even if in fact this page has little or nothing to do with the anchor text. Importing the related bit vector from a target document URL2 inho the phrase A related phrase bit vector for document URLO eliminates the reliance of the search sysfiem on just the relationship of phrase A in tTRLO pointing to URL1 as an indicator of significance or iTRLl to the anchor text phrase.
(0149j Each phrase in the index 15Q is also given a phrase number, based on its frequency of occurrence in the coxpus. The more common the phrase, the Lower phrase number zt receivesorder in the index. The indexing system 120 then sorts 506 alI of the posting lists in the index 150 in declining order according to the number of documents listedphrase number of in each posting List, so that the most frequently occurring phrases are listed first. The phrase number can then be used to look up a particular phrase.
IIL Search S, sy tem 10150] The search system 120 operates to receive a query and search far documents relevant to the query, and provide a list of these documents (with links to the documents) in a set of search xesults. FIG. 6 illustrates the main functional operations of the search system 120:
[0151] 600: Identify phrases in the query.
(0152) 602: Retrieve documents relevant to query phrases.
[0153j 604: Rank documents in search results according tv phrases.
(0154] The details of each of these of these stages is as follows.
1. Identification of Phrases in the Quey and Quer3r Expansion (0155] The first stage 600 of fine search system 120 is to identify any phrases that are present in the query in order to effectively search the index. The following terminology is used in this section:
[d156] q: a query as input and recezve by the search system 120.
(0157] Qp: phrases present in the query.

j0158] Qr: related phrases of Qp.

j0159] Qe: phrase extensions of Qp.

[0160] Q: the union of Qp and Qr.

[0161] A query q is received from a client 190, having up to some ma~amum number characters or words.
:of [0T.62] A phrase window of size N (e.g., 5~ is used by the search system 120 ~
traverse the terms of the query q. The phrase window starts with fine first term of the query, eictends N terms to the right This window is then shifted right M N
times, where M is the number of terms in the query.
[01b3] At each window position, there will be N terms (or fewer) fiern~s in the window. These terms constitute a possible query phrase. The possible phrase is looked up in the good phrase list 208 to determine if it is a good phrase or not. If the possible phrase is present in the good phrase List 208, then a phrase number is rid far phrase; the possible phrase is now a candidate phrase.
r0164] After all possible phrases in each window have been tested to determine if they are good candidate phrases, the search system 120 will have a set of phrase numbers for the corresponding phrases in the query. These phrase numbers are then sorbed (declining order).
[0165) Starting with the highest phrase number as the first candidate phrase, the search system 1Z0 determines if there is another candidate phrase within a fixed numerical distance within the sorted list, i.e., the difference between the phrase numbers is within a threshold amount, e.g. 2(?,00(?. If so, then the phrase that is leftmost in the query is selected as a valid query phrase Qp. This query phrase and all of its sub-phrases j~s removed from the list of candidates, and the list is resorted and the process repeated. The result of this process is a set of valid query phrases Qp.
[4166] For example, assume the search query is "Hillary Rodham Clinton Bill on the Senate Floor". The search system 120 would identify the following candidate phrases, "H4lary Rodham Clinton Br71 on," "Hillary Rodham Clinton Bill," and "Hivary Rodharn Ctinton". The first two are discarded, and the last one is kept as a valid query phrase. :Next the search system 120 would identify "Bill on the Senate Floor", and the subsphr~ses "Bill on the Senate", "Bill on the', "Bill on", "Bill", and would select "Bill"
as a valid query phrase Qp. Finally, the search system 120 would parse "on the senate floor" axed identify "Senate Floor" as a valid query phrase.
]0167] Next, the search system 120 adjusts the valid phrases Qp for capitalization. When parsing the query; the search system 120 identifies potential capitalizations in each valid phrase. This may be done using a table of known capitalizations, such as "orated states" being capitalized as "United States", or by using a grammar based capitalization algorithm. This produces a set of properly capitalized query phxases.
[0168] The search system 120 then makes a second pass through the capitalized phrases, and selects only those phrases are leftmost and capitalized where both a phrase and its subphrase is present in the set. For example, a search on "president of the united states" will be capitalized as "President of the United States" .
[0169] In the next stage, the search system 120 identifies 602 the documents that are relevant to the query phrases Q. The search system 120 then retrieves the posting lists of the query phrases Q, and intersects these lists to determine which documents appear on the all (or some number) of the posting lists for the query phrases.
If a phrase Q in the query has a set of phrase extensions Qe (as further explained below), then the search s~s~~m 120 first forms the union of the posting lists of the phrase extensions, prior to dying; the intersection with the posting lists. The search system 120 identifies phrase extensions by looking up each query phrase Q in the incomplete phrase list 216, as described above.
(017Uj The result of the intersection is a set of documents that are relevant to the query. liidexing docrxments by phrases and related phrases, identifying phrases Q in the query, and then expanding the query to include phrase .extensions results in the selective of a set of documents that are mere relevant to the query then would result in a conventional Boolean based search system in which only documents that contain the query terms are selected.
(OI7I) In one embodiment, the search system 120 can use an optinZized mechanism to identify documents responsive to the query without having to intersect all of the posting lists of the query phrases Q. As a result of the structure of the index 150, for each phrase g~, the related phrases gk are known and identified in the related phrase bit vector for gx. Accordingly, this information can be used to shortcut the intersection process where two or more query phrases are related phrases to each other, or have common related phrases. In those cases, the related phrase bit vectors can be directly accessed,, and then used next to rehieve corresponding documents. This process is more fully described asfvllows.
(Q172) Given any fwv query phrases Ql and Q2, there are three possible cases of relations:
(0173) T) Q2 is a related phrase of Q1;

j0174I 2) Q2 is not a related phrase of Q1 and their respective related phrases Qr1 anii Qr2 do not intersect (i.e., no common related phrases) and [0175] 3) Q2 is not a related phrase of Q1, but their respective related phrases Qrl ansi Qr2 do intersect.
[017Gj For each pair of query phrases the seaxch system 320 determines the appropriate case by looking up the relatred phrase bit vector of the query phrases Qp.
[0177] The search system 120 proceeds by retrieving the posting list for query phrase:Ql, which contains the documents containing Q1, and for each of these documents, a related phrase bit vector. The related phrase bit vector for Q1 will indicated whether phrase Q2 (and each of the remaining query phrases, if any) is a related phrase of Q1 and is present in the document.
[OT78~ If the first case applies to Q2, the search system 120 scans the related phrase bit nectar for each document d in Q1's posting list to determine if it has a bit set for Q2. If this bit is.not set in for document d in Ql's posting list, then it mans that Q2 does nvt appear in that document. As result, this document can be immediabeIy eliminated from further consideration. The remaining documents can then be scored.
This means further that it is vnne~cessary for the search system T20 to process the posting lists of QZ to see which documents it is present in as well, thereby saving compute time.
(OT79~ Tf the second case applies to Q2, then the two phrases are unrelated to each other. For example the query "cheap bolt action rifle" has two phrases "cheap" and "bolt action rifle". Neither of these phrases is related to each other, and further the related~phrases of each of these do not overlap; i.e., "cheap" has related phrases "low CoSt," "IIleXpenSlVe," "dlSCOUnt," "baIgaln baSeIIlent," and "lOtlSy,", Whereas "bolt action rifle" has related phrases "gun,' "2Z caliber", "magazine fed," and "Armalite AR

3flM", ~~hich lists thus do not intersect. In this case, the search system 220 does the regular intersection of the posting lists of Q? and Q2 to obtain the documents ~or scoring.
[184] If the third case applies, then here tire two phrases Ql and Q2 that are not related, but that do have at least one related phrase in common. For example the phrases "bolt action rifle" and "22" would both have "gun" as a related phase. In this case, the search system 120 retrieves the posting lists of both phrases Q? and Q2 and intersects the lists to produce a list of documents that contain both phrases.
[0181] The search system 120 can then quickly score each of the resulting documents. First, the search system 1Z0 determines a score adf ustment value for each document. The score adjustment value is a mask formed from the bits in dte positions corresponding to the query phrases Q1 and Q2 in the related phrase bit vector for a document. For example, assume that Q? and Q2 correspond to the 3~ and 6~ bi-bit positions in the related phrase bit vector far document d, and the bit values in 3~
position are (1,1j and the bit values in the 6~ pair are (?,0), then the score adjustment value is the bit mask "00 0011 00 0010". The score adjustment value is then used to mask the related phrase bit vector for the documents, and modified phrase bit vectors then are passed into the ranking function (next described) to be used in calculating a body score for the documents.
2.
a) Ranking Documents Based on Contained F~rases 0182] The search system 12U provides a rankiung stage 604 in which the documents in fine search results are ranked, using the phrase information in each document's related phrase bit vector, and the cluster bit vector for the query phrases.

This approach ranks documents according to the phrases that are contained in the document or informally "body hits.'.
[01$3] As described above, for any given phrase g;, each document d in the g~'s posting Iist has an associated related phrase bit vector that identifies which related phrases gk and which secondary related phrases g: are present in document rI.
The more related:phrases and secondary related phrases present in a given document, the more bits that will be set in the document's related phrase bit vector for the given phrase. The more bits that are set, the greater the numerical value of the related phrase bit vector.
[01$4] Accordingly, in one embodiment, the search system 120 sorts the documents irt the search results according to the value of their related phrase bit vectors.
The documents containing the most related phrases to the query phrases Q will have the highest valued related phrase bit vectors, and these documents will be the highest ranking documents in the search results.
[0185] This approach is desirable because semantically, these documents are most topically relevant to the query phrases. Note that this approach provides highly relevant documents even if the documents do not contain a high frequency of the input query terms q, since related phrase information was used to both identify relevant documenfis, and then rank these documents. Documents with a low frequency of the input query terms may still have a large number of related phrases to the query terms and phrases and thus be more relevant than documents that have a high frequency of just the-query terms and phrases but no related phrases, [0185] In a second embodiment, the search system 120 scares each document in the result set according which related phrases of the query phrase Q it contains. This is done as follows:

[D187( Given each query phrase Q, there will be some number N of related phrases Qr to the query phrase, as identified during the phrase identification process.
As did above, the related query phrases Qr are ordered acrnrding to their information gain from the query phrase Q. These related phrases are then assigned points,=started with N points for the First related phrase Qr1 [i.e., the relatied phrase Qr with thje highest information gain from Q), then N 1 points for the next related phrase Qr2, then N 2 points for Qr3, and so on, so that the last related phrase QrN
is assigned 1 point. .
[0188] Each document in the search results is then scored by determining which related phrases Qr of the query phrase Q are present, and giving the document the points assigned to each such related phrase Qr. The documents are then sorbed from highest to lowest score.
[D189j As a further refinement, the search system 120 can cull certain documents from the result set In some cases documents may be about many different topics; this is particularly the case for longer documents. In many cases, users prefer documents that are stroi<gly on point with reapecfi to a single topic expressed in the query over documents that are relevant to many different topics.
[019Dj To cull these latter types of documents, the search system 120 uses the cluster information in the cluster bit vectors of the query phrases, and removes any document in which there are more than a threshold number of clusters in the document For example, the search system 120 can remove any documents that contain more than two clusters. This cluster threshold can be predetermined, or set by the user as a search parameter.

b) ~ Documents Based on Anchor Phrases (0191] In addition to ranking the documents in the search results based on body hits of query phrases Q, in one embodiment, the search system 120 also ranks the documents based on the appearance of query phrases Q and related query phrases Qr in anchor6 to other documents. In one embodiment, the search system 120 calculates a score fQr each docwnent that is a fiwctaon (e.g., linear combination) of two scores, a body hit score and an anchor hit score.
(0192] For example, the document score for a given document can be calculated as follows:
j01931 Score ~ .34*(body hit score)+.70*(anchor hit score).
(~)194j The weights of .30 and .70 can be adjusted as desired. The body hit score for a document is the numerical value of the highest valued related phrase bit vector for the document, given the query phrases Qp, in the manner described above.
Alternatively, this value can directly obtained by the search system I20 by looking up each query phrase Q in the index 250, accessing the document from the posting fist of the query phrase Q, and then accessing the related phrase bit vector.
[0195] The anchor hit score of a document d a function of the related phrase bit vectors' of the query phrases Q, where Q is art anchor term in a document that references document d. When the indexing system 110 indexes the documents in the document collection, it maintains for each phrase a list of the documents in which the phrase is anchortext m an outlink, and also for each document a list of the in)inks (and the associated anchor text) from other documents. The iniinks for a document are references (e.g. hyperlinks) from other documents (referencing docwments) to a given document.

(0196] To determine the anchor hit score for. a given document d then, the search systemi120 iterates over the set of referencing documents R (i=1 to number of referencing documents) listed in index by their anchor phrases Q, and sums the following product:
0197] P~,-.Q.Related phrase bit vector'~D.Q.ReIated phrase bit vector.
[019$] The product value here is a score of how topical anchor phrase Q is to document D. This score is here called the "inbound score. component:" This product effectively weights the current dorunnent D's related bit vector by the related bit vectors of anchor phrases in the referencing document R. If the referencing documents R
themselves are related to the query phrase Q (and thus, have a higher valued related phrase bit vector), then this increases the significance of the current docament D score.
The body hit score and the anchor hit score are then combined to create the document score, as described above.
[0199] Next, for each of the referencing documents R, the related phrase bit vector for each anchor phrase Q is obtained. This is a measure of haw bopicaI
the anchor phrase Q is to the document R. This value is here called the outbound score component.
[0200] From the index 1~0 then, all of the (referencing document, referenced document) pairs are extracted for the anchor phrases Q. These pairs are then sorted,by their associated (outbound srnre component, inbound score component) values.
Depending on the implementation, either of these components can be the primary sort key, and the other can be the secondary sort key. The sorted results are then presented to the user. Sorting the documents on the outbound score component makes documents that have many related phrases to the query as anchor hits, rank most highly, thus represeinting these documents as "expert" documents. Sorting on the inbound document score makes documents that frequently referenced by the anchor ternLS
the most high ranked.
3. Phrase Based PersonaIiza~ion of Search [U2p~] Another aspect of the search systenn 120 is the capab~.ity to personalize 606 or customize the ranking of the search results in accordance with a model of the usef s particular interests. In this manner, documents that more likely to be relevant to the user's interests are ranked higher in the search results. The personalization of search result is as follows.
[0202] As a preliminary matter, it is useful to define a user's interests ~e.g., a user model) in terms of queries and documents, both of which can be represented by phrases.
For an input search query, a query is represented by the query phrases Q, the related phrases of Qr, and phrase extensions Qe of the query phrases Qp. This set of terms and phrases thus represents ~e meaning of ~e query. Next, the meaning of a document is represented by the phrases associated with the page. As described above, given a query and document, the relevant phrases for the document are determined from the body scores (the related bit vectors) for all phrases indexed to the document.
Finally, a user can be represented as the union of a set of queries with a set of documents, in terms of the phrase;9 that represent each of these elements. The particular documents to include in the set representing the user can be determined from which documents the user selects in previous search results, or in general browsing of the corpus (e.g., accessing documents on the Internet), using a client side tool which monitors user actions and destinations.
[0203) The process of constructing and using the user model for personalized ranking is as follows.

j0204] First, for a given user, a Iist of the last K queries and P documents access~id is maintained, where K and P are preferably about 250 each. The lists may be maintained in a user account database, where a user is recognized by a login or by browses cookies. For a given user, the lists will be ennpty the first time the user provides a query.
[0205] Next ~a query q is received from the user. The related phrases Qr of q are retrieved, along with the phrase extensions, in the manner described above.
This forms the query model.
j020b] In a first pass (e.g., if there are no stored query information for the user), the search system 12a opexates to simply return the relevant documents in the seaxch result to the nse~'s query, without further customized ranking.
[0207] A client side browses tool morubors which of the documents in the search results the user accesses, e.g., by clicking on the docunnent link in the search results.
These accessed documents for the basis for selecting which phrases will become part of the user model. For each such accessed document, the search system 120 retrieves the document model for the document, which is a list of phrases related to the document.
Each phrase that is related to the accessed document is added to the user model.
[020$J Next, given the phrases related to an accessed document, the clusters associated with these phrases can be determined from the clusier bit vectors for each phrase.-For each cluster, each phrase that is a member of the cluster is determined by looking the phrase up in its related phrase table that contains the cluster number, or cluster bit vector representation as described above. This cluster number is them added to the user model. In addition, for each such cluster, a counter is maintained and incremented each time a phrase in that cluster is added to the user model These counts may be lured as weights, as described below. Thus, the user model is built fronn phrases included in clusters that are present on a document that the user has expressed an interest'in by accessing the document.
[0209] The same general approach can be more precisely focused to capture phrase information where a higher level of interest than merely accessing the document is manifested by the user (which the user may do simply to judge if indeed the doc~mnent is relevant). For example, the collection of phrases into the user model may be limited to those documents that the user has printed, saved, stored as a favorite or link, entail to another user, or maintained open in a browser window for an extended period of time (e.g.,10 minute's). These and other actions manifest a higher level of interestin the document.
[0210] When another query is received from the user, the related query phrases tar are retrieved. These related query phrases (fir are intersected with the phrases listed in the user model to determine which phrases are present in both the query and the user model. A mask bit vector is initialized for the related phrases of the query Qr. This bit vector is a ti bit vector as de~n'bed above. For tech related phrase Qr of the query that is also present in the user model, both of the bits for this related phrase are set in the mask bit vector. The mask bit vector thus represents the related phrases present in both the query and the user model.
[0211] The mask bit vector is then used to mask the related phrase bit vector for each dacunnent in the current set of search results by ANDirtg the related phrase bit vector with the mask bit vector. This has the effect of adjusting the body score and the anchor hit score by the mask bit vector. The documents are then scored for their body score and anchor score as before and presented to the user. This approach essentially requires that a document have the query phrases fat are included in the user model in order t4 be highly ranked.
(d212j As an alternative embodiment, which does not impose the foregoing tight constraint, the mask bit vector can be cast into array, so that each bit is used to weight the duster counts for the related phrases in the user model. Thus, each of the cluster counts gets multiplied by 0 or 1, effectively zeroing or maintaining the counts. Next, these counts themselves are used as weights are also used to multiply the related phrases for each document chat is being scored. This approach has the benefit of allowing documents that do not have the query phrases as related phrases to still score appropriately.
[0213] Finally, the user model may be limited to a current session, where a session:is an interval of time for active period of time in search, after which session the user model is dumped. Alternatively, the user model for a given user may be persisted over time, and then dawn-weighted or aged.
1V. Result Presentation (02I4j The presentation system 130 receives the scored and sorted search results from the search system 120, and performs further organizational, annotation, and clusterisig operations prior to presenting the results to the user. These operations facilitatie the user's understanding of the content of the search results, eliminate duplicates, and provide a more representative sampling of the search results.
FTG. 7 illusixates the main functional operations of the presentation system 130:
[0215j 700: Quster documents according to topic clusters [0216j 702: Generate document descriptions 4b [02I7j ' 704: Eliminate duplicate documents.
[0218] Each of these operations takes as an input the search results 701 and outputs modified search results 703. As suggested by FIG. ~, the order of these operations is independent, and may be varied as desired for a given embodiment, and thus the inputs may be pipelined instead of being in parallel as shown.
I. Dynamic Taxonomy Generation for Presentation [0219] For a given query, it is typical to return hundreds, perhaps even thousands of documents that satisfy the query. In many cases, certain documents, while having different content from each other, are sufficiently related to form a meaningful group of related document, essentially a cluster. Most users however, do not review beyond the first 30 or 40 documents in the search results. Thus, if the first documents for example, would come from three clusters, but the next 10D
documents represent an additional four dusters, then without further adjustment, the user will typically not review these later documents, which in fact may be quite relevant to the user's query since they represent a variety of different topics related to the query. Thus, it is here desirable to provide the user with a sample of documents from each cluster, thereby exposing the user to a broader selection of different documents from the search results. The presentation system 130 does this as follows.
[0220] As in other aspects of the system 10U, the presentation system 130 makes use of the related phrase bit vector for each document d in the search results. More specifically, for each query phrase Q, and for each d~ument d in Q's posting list, the related phrase bit vector indicates which related phrases Qr are present in the document.
Over the set of documents in the search results then, for each related phrase Qr, a count is dehermined for how many documents contain the related phrase Qr by adding up the bit values in the bit position corresponding to Qr. When summed and sorted over the search results, the most frequently occurring related phrases Qr will be indicated, each of which will be a cluster of documents. The most frequently occurring related phrase is the first; cluster, which takes as its name its related phrase Qr, and so on for the top three to five clusters. Thus, each of the top clusters has been identified, along with ii~e phrase Qr as a name or heading for the cluster.
[0221] Now, documents from each cluster can be presented to the user in various:ways. In one application, a ~lxed number of documents from each cluster can be presented, far example, the 10 top scoring documents in each cluster. In another applicafion, a proportional number of documents from each cluster may be presented.
Thus, ifahere are 100 documents in the search result, with 50 in cluster 1, 30 in cluster 2, in cluster 3, 7 in cluster 4, and 3 in cluster 5, and its desired to present only ZO
documents, then the documents would be select as follows: 10 documents from cluster 1; 7 documents from cluster 2, 2 documents from clustex 3, and 1 document from cluster ~. The docuanents can then be shown to the user, grouped accordingly under the appropriate cluster names as headings.
[0222] For example, assume a search query of "blue merle agility training", for which the search system 120 retrieves 100 documents. The search system 120 will have already.identified "blue merle' and "agMity training" as query phrases. The related phrases of these query phrases as:
[0?.23] "blue merle":: "Australian Shepherd," "red Irierle," "'fI'1C0101,'~
aallSSle"i [0224] "agility training':: "weave poles," "teeter," "tunnel," "obstacle,"
"border collie'.
4$

[0225) The presentation system 130 then determines for each of the above related phrases of each query phrase, a count of the number of documents contain such phrase. For example, assume that the phrase "weave poles" appears in 75 of the documents, "teeter" appears in 60 documents, "red merle" appears in 50 documents, Then the first cluster is named "weave poles" and a selected number of documents from that cluster are presented; the second cluster is named uteeter," and selected number are presented as well, and so forth. For a fixed presentation, 20 documents from each cluster may be selected. A proportional presentation would use a proportionate number of documents from each cluster, relative to the total number of documents.
2. Topic Based Document Descriptions [0226] A second function of the presentation system 130 is the creation 702 of a document description that can inserted into the search result pxesentation for each document These descriptions are based an the related phrases that are present in each document, and thus help the user understand what the document is about in a way that is contextually related to the search. The document descriptions can be either general or perso~n~lized to the user.
a) General Topic Document Descri~tians [0227) As before, given a query, the search system 120 has determined the related query phrases Qr and the phrase extensions of the query phrases as well, and then identified the relevant documents for the query. The presentation system accesse~ each document in the search results and perform the follow operations.
[02281 First, the presentation system 130 ranks the sentences of the document by the number of instances of query phrases Q, related query phrases Qr, and phrase extensions Qp, thereby maintaining far each sentence of a document counts of these three aspects.
[0229] Then the sentence are sorted by these counts, with the first sort key being the count of query phrases Q, the second sort key being the count of related query phrases Qr, and the final sort key being the count of phrase extensions Qp.
[0230] Finally, the top N (e.g., 5) sentences following the sort are used as the description of the document, This set of sentences can be formatted and included in the presentation of the document in the modified search results 703. This process is repeated for some number of documents in the search results, and may be done on demand each time the user requests a next page of the results.
b) Personalnzed Topic Based Document Descriptions [0231] In embodiments where personalization of the search results is provided, the document descriptions can likewise be personalized to reflect the user interests as expressed in the user model. The presentation system 130 does this as follows.
[0232] First, the presentation system 230 determines, as before, the related phrases that are relevant to the user by intersecting the query related phrases Qr with the usei model (which lists the phrases occurring in documents accessed by the user).
10233] The presentatian systenn 130 then stable sorts this set of user related phrases Ur according to the value of the bit vectors ~emseIves, prepending the sorted list to the list of query related phrases Qr, and removes any duplicate phrases. The stable sort maintains the existing order ~of equally ranked phrases. This results in a set of related phrases which related to the query or the user, called set øu.
[0234] Now, the presentation system 130 uses this ordered list of phrases as die basis for ranking the sentences in each document in the search results, in a manner similar to the general document description process described above. Thus, for a given document, the presentation system 130 ranks the sentences of the document by the number of instances of each of the user related phrases and the query related phrases Qu, and sorts the ranked sentences according to the query counts, and finally sorts based on the number of phrase extensions for each such phrase. Whereas previously the sort keys where in the order of the query phrases Q related query phrases C,~r, and phrase extension Qp, here the sort keys are in the order of the highest to lowest ranked user related phrases Ur.
[023a~ Again, this process is repeated for the documents in the search results (either ~n dem nd or afarehand). For each such document then the resulting document description comprises the N top xanlced sentences from the document. Here, these sentences will the ones that have the highest numbers of user related phrases Ur, and bus represent the key sentences of the docmment that express the concepts and topics most relevant to the user (at least according to the information captured in the user model).
3. Duplicate Document Detection and Elimination [0236 In large corpuses such as the Internet, it is quite common for there to be multiple instances of the same document, or portions of a document in many different locations. For example, a given news article produced by a news bureau such as the Associated Fress, may be replicated in a dozen or more websites of individual newspapers. Including all of these duplicate documents in response to a search query only burdens the user with redundant information, and does not usefully respond to the query. Thus, the presentation system 130 provides a further capability 704 to identify documents that are likely to be duplicates or near duplicates of each other, and only include; one of these in the search zesults. Consequently, the user receives a much more diversified and robust set of results, and does not have to waste time reviewing documents that are duplicates of each othex. The presentation system 130 provides the functioxzality as follows.
(0237] The presentation system 130 processes each document in the search result set 701. For each document d, the presentation system 130 first determines the list of related phrases R associated with the document. For each of these related phrases, the presentation system 130 ranks the sentences of the document according fio the frequency of occurrence of each of these phrases, and then selects the bop N (e.g., 5 to 10) ranking sentences. 'This set of sentences is then stored in association with the document. One way bo do this is to concatenate the selected sentences, and then take use a hash table to store the document identifier.
j0238) Then, the presentation system 130 compares the selected sentences of each document d to the selected sentences of the other documents in the search results 701, and if the selected sentences match (within a tolerance), the documents are presumed to be duplicates, and one of them is removed from the search results.
For example, the presentation system 130 can hash the concatenated sentences, and if the hash table already has an entry for the hash value, then this indicates that the current document and presently hashed document are duplicates. The presentatian system can then update the table with the document ID of one of the documents.
Preferably, the presentation system 130 keeps the document that has a higher page rank or other query izadependent measure of document significance. In addition, the presentation system 130 can modify the index 150 to remove the duplicate document, so that it will not appear in future search results for any query.

j0289] The same duplicatie elimination process may be applied by the indexing system X20 directly. When a document is crawled, the above described document I
description process is performed to obtain the selected sentences, and then the hash of these sentences. If the hash table is filled, then again the newly crawled document is deemed to be a duplicate of a previous document. Again, the indexing system I10 can then keep the document with the higher page rank or other query independent.measure.
[a240j The present invention has been described in particular detail with respect to one possible embodiment. Those of skill in the art will appreciate that the invention may be practiced in other embodimenfis. First, the particular naming of the components, capitalization of terms, the attributes, data strictures, or any other programming or structural aspect is not mandatory or significant, and the mechanisms that implement the invention or its features may have different names, formats, or protocols.
Further, the system may be implemented via a combination of hardware and software, as described, or entirely in hardware elements. Also, the particular division of functionality between the various system components described herein is merely exemplary, and not mandatory; functions performed by a single system component may instead be performed by multiple components, and functions performed by multiple componenfis may instead performed by a single component.
~0241j Some portions of above description present the features of the present invention in terms of algorithms and symbolic representations of operations on informaition. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work ta~others skilled in the art. These operations, while described functionally or logically, are understood to be implemented by computer programs. Furthermore, it has also prQVen convenient at times, to refer to these arrangements of operations as modules or by functional names, without loss of generality.
[0242) ~ Unless specifically stated otherwise as apparent from the above discussion, it is appreciated that throughout the description, discussions utilizing terDns such as "processing" ~or "computing" or "calculating" or "determining' or "displaying"
or the like, refer to the action and processes of a computer system, yr similar electronic computing device, that manipulates and transforms data represented as physical electronic) quantities within the computer system mennories or registers or other such information storage, transmission or display devices.
[0243] Certain aspects of the present invention include process steps and instructions described herein in the form of an algorithm. It should be noted that the process. steps and instructions of the present invention could be embodied in software, firmwaie or hardware, and when embodied in software, could be downloaded to reside on and be operated from different platforms used by real time network operating systems.
(0244] The present invention also relates to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored on a computer xeadable medium that can be accessed by the computer. Such a computer program may be stored in a computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMS), EPROMs, EEPROMs, magnetic ar optical cards, application specific integrated circuits (ASICs), or any type of media suitable for storing electronic instructions, and each coupled do a computer system bus.
FurtherJnore, the computers referred to in the specification may include a single procession or may be architectures employing multiple processor designs for increased computing capability.
[0245] The algorithms and operations presented herein are not inherently related to any particular computer or other apparatus. Various general purpose systems may also be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will be apparent to those of skill iut the, along with equivalent variations. In addition, the present invention is not drescribed with reference to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any references to specific languages are provided for disclosure of enablement and best mode of the present invention.
[0246] The present invention is well suited to a wide variety of computer network systems over numerous topologies. Within this field, the configuration and management of large networks comprise storage devices and computers that are communicatively coupled to dissimilar computers and storage devices over a network, such as the Internet.
[0247] Finally, it should be noted that the language used in the specification has been principally selected for. readability and instructional purposes, and may not have been selected to delineate or circumscribe the inventive subject matter.
Accordingly, the disclosure of the present invention is intended to be illustrative, but not limiting, of the scope of the invention, which is set forth in the following claims.
~b

Claims (26)

1. ~A computer implemented method for identifying valid phrases in a document collection, the method comprising:
collecting possible phrases from documents in the document collection;
classifying each possible phrase as either a good phrase or a bad phrase according to frequency of occurrence of the possible phrase; and selectively retaining only good phrases that predict the occurrence of at least~
one other good phrase in the document collection.
2 ~The method of claim 1, wherein collecting possible phrases comprises:
traversing the words of a document with a multiword phrase window, and selecting as candidate phrases all sequences of words in the window that begin wig a first word in the window.
3. ~The method of claim 2, wherein the phrase window includes at least 4 words.
4. ~The method of claim 1, wherein collecting possible phrases comprises:
maintaining for each possible phrase and each good phrase a frequency count of the number of documents containing the phrase;
maintaining for each possible phrase and each good phrase a frequency count of the number of instances of the phrase; and maintaining for each possible phrases and each good phrase a frequency count of the number of distinguished instances of the phrase.
5. ~The method of claim 4, wherein a distinguished instance of a phrase comprises a phrase distinguished from neighboring content in the document by grammatical or format markers.
6. ~The method of claim 2, wherein classifying each possible phrase as either a good phrase or a bad phrase comprises:
classifying a possible phrase as a good phrase where the possible phrase appears in a minimum number of documents, and appears a minimum number of instances in the document collection.
7. ~The method of claim 1, wherein classifying each possible phrase as either a good phrase or a bad phrase comprises:
classifying a possible phrase as a good phrase where the possible phrase appears in a minimum number of distinguished instances in the document collection.
8. ~The method of claim 1, wherein selectively retaining good phrases that predict the occurrence of at least one other good phrase in the document collection comprises:
retaining the good phrase only if an information gain of the good phrase with respect to at least one other good phrase exceeds a threshold greater than 1.
9. ~The method of claim 8, wherein the information gain of a good phrase gj with respect to another good phrase g k is:
I(jk) = A(j,k)/E(j,k) where A(j,k) is an actual co-occurrence rate of gj and gk; and E(j,k) is an expected co-occurrence rate gj and gk.
10. The method of claim 9, wherein good phrases gj and gk co-occur in a document when gk and gk are within a predetermined number of words of each other.
11. The method of claim 1, wherein retaining good phrases that predict the occurence of at least one other good phrase in the document collection comprises:
removing a good phrase having an information gain with respect to a plurality of other good phrases less than a predetermined threshold.
12 ~The method of claim 1, further comprising:
removing incomplete phrases from the good phrases.
13. The method of claim 12, wherein an incomplete phrase is a phrase that only predicts its phrase extensions, and wherein a phrase extension of a phrase is a super-sequence of the phrase that begins with the phrase.
14. The method of claim 12, further comprising:
maintaining for each incomplete phrase at least one phrase extension of the~
incomplete phrase; and responsive to a phrase in a search query being an incomplete phrase, including in the search query at least one phrase extension of the incomplete search phrase.
15. The method of claim 1, further comprising:
determining co-occurrence counts for co-occurring good phrases; and determining whether a good phrase gj predicts another good phrase gk as a function of a co-occurrence rate of co-occurrences of gj, gk, and an expected rate of co-occurrences of gj and gk.
16. The-method of claim 1, further comprising:
identifying for a good phrase, at least one other good phrase that is a related phrase of the good phrase.
17. The method of claim 16, wherein a good phrase gj is a related phrase of another good phrase gk where the information gain of gj with respect to gk exceeds a predetermined threshold.
18. A method for identifying related phrases in a document collection, the method comprising:
determining for each of a plurality of phrases a frequency of occurrence of the phrase in the document collection;
determining for each of a plurality of pairs of the phrases, a co-occurrence rate of the pair of phrases in the document collection;
determining for a pair of phrases gj and gk in the document collection an information gain of phrase gk with respect to gj as a function of the co-occurrence rate of gj and gk, and the frequencies of gk and gj in the document collection; and identifying gk as a related phrase of gj where the information gain of gk in the presence of gj exceeds a predetermined threshold.
19. The method of claim 18, wherein the predetermined threshold is about 100.
20. The method of claim 18, wherein the information gain I of gk in the presence of gk is I(jk) = A(j,k)/E(j,k) where A(j,k) is an actual co-occurrence rate of gj and gk; and E(j,k) is an expected co-occurrence rate gj and gk.
21. The method of claim 18, further comprising:
for each phrase gj, identifying a cluster comprising the phrase and at least one related phrase gk.
22. The method of claim 18, further comprising:
for each phrase gj, identifying a set R including a plurality of related phrases;
determining for each pair of related phrases in set R an information gain of the pair of related phrases; and identifying as a cluster of related phrases of gj, phrase gj, and each the related phrase in set R that has non-zero information gain with respect to each other phrase in set R.
23. The method of claim 22, further comprising:
assigning each cluster a unique cluster number as a function of the related phrases included in the cluster.
24. The method of claim 22, further comprising:
assigning to the cluster a name comprising the related phrase having a highest information gain of the related phrases in the cluster.
25. The method of claim 18, further comprising:
for a phrase gj, storing a bit vector in which each bit position corresponds to a valid phrase, and the bit corresponding to the bit position indicates whether the valid phrase is a related phrase of gj.
26. A computer program product, for identifying valid phrases in a document collection, comprising computer operable instruction stored on a computer accessible medium and adapted to control a processor to perform the operation of:
collecting possible phrases ham documents in the document collection;
classifying each possible phrase as either a good phrase or a bad phrase according to frequency of occurrence of the possible phrase; and selectively retaining only good phrases that predict the occurrence of at least~
one other good phrase in the document collection.
CA2513850A 2004-07-26 2005-07-26 Phrase identification in an information retrieval system Expired - Fee Related CA2513850C (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US10/900,021 2004-07-26
US10/900,021 US7580921B2 (en) 2004-07-26 2004-07-26 Phrase identification in an information retrieval system

Publications (2)

Publication Number Publication Date
CA2513850A1 true CA2513850A1 (en) 2006-01-26
CA2513850C CA2513850C (en) 2014-01-14

Family

ID=34941843

Family Applications (1)

Application Number Title Priority Date Filing Date
CA2513850A Expired - Fee Related CA2513850C (en) 2004-07-26 2005-07-26 Phrase identification in an information retrieval system

Country Status (11)

Country Link
US (3) US7580921B2 (en)
EP (1) EP1622053B1 (en)
JP (1) JP4976666B2 (en)
KR (1) KR101190230B1 (en)
CN (1) CN1728142B (en)
AT (1) ATE500559T1 (en)
AU (1) AU2005203240B2 (en)
BR (1) BRPI0503781A (en)
CA (1) CA2513850C (en)
DE (1) DE602005026609D1 (en)
NO (1) NO20053638L (en)

Families Citing this family (113)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7266553B1 (en) * 2002-07-01 2007-09-04 Microsoft Corporation Content data indexing
US7574016B2 (en) * 2003-06-26 2009-08-11 Fotonation Vision Limited Digital image processing using face detection information
US7617205B2 (en) 2005-03-30 2009-11-10 Google Inc. Estimating confidence for query revision models
US9223868B2 (en) 2004-06-28 2015-12-29 Google Inc. Deriving and using interaction profiles
US7580921B2 (en) 2004-07-26 2009-08-25 Google Inc. Phrase identification in an information retrieval system
US7711679B2 (en) 2004-07-26 2010-05-04 Google Inc. Phrase-based detection of duplicate documents in an information retrieval system
US7567959B2 (en) 2004-07-26 2009-07-28 Google Inc. Multiple index based information retrieval system
US7536408B2 (en) 2004-07-26 2009-05-19 Google Inc. Phrase-based indexing in an information retrieval system
US7580929B2 (en) * 2004-07-26 2009-08-25 Google Inc. Phrase-based personalization of searches in an information retrieval system
US7599914B2 (en) * 2004-07-26 2009-10-06 Google Inc. Phrase-based searching in an information retrieval system
US7584175B2 (en) 2004-07-26 2009-09-01 Google Inc. Phrase-based generation of document descriptions
US7702618B1 (en) 2004-07-26 2010-04-20 Google Inc. Information retrieval system for archiving multiple document versions
US7199571B2 (en) * 2004-07-27 2007-04-03 Optisense Network, Inc. Probe apparatus for use in a separable connector, and systems including same
US7962510B2 (en) * 2005-02-11 2011-06-14 Microsoft Corporation Using content analysis to detect spam web pages
US7870147B2 (en) * 2005-03-29 2011-01-11 Google Inc. Query revision using known highly-ranked queries
US7765214B2 (en) * 2005-05-10 2010-07-27 International Business Machines Corporation Enhancing query performance of search engines using lexical affinities
US9110985B2 (en) 2005-05-10 2015-08-18 Neetseer, Inc. Generating a conceptual association graph from large-scale loosely-grouped content
US7958120B2 (en) 2005-05-10 2011-06-07 Netseer, Inc. Method and apparatus for distributed community finding
US8321198B2 (en) * 2005-09-06 2012-11-27 Kabushiki Kaisha Square Enix Data extraction system, terminal, server, programs, and media for extracting data via a morphological analysis
US7971137B2 (en) * 2005-12-14 2011-06-28 Google Inc. Detecting and rejecting annoying documents
WO2007084616A2 (en) * 2006-01-18 2007-07-26 Ilial, Inc. System and method for context-based knowledge search, tagging, collaboration, management and advertisement
WO2007084778A2 (en) * 2006-01-19 2007-07-26 Llial, Inc. Systems and methods for creating, navigating and searching informational web neighborhoods
US8229733B2 (en) * 2006-02-09 2012-07-24 John Harney Method and apparatus for linguistic independent parsing in a natural language systems
US8843434B2 (en) * 2006-02-28 2014-09-23 Netseer, Inc. Methods and apparatus for visualizing, managing, monetizing, and personalizing knowledge search results on a user interface
US8126874B2 (en) * 2006-05-09 2012-02-28 Google Inc. Systems and methods for generating statistics from search engine query logs
US9646089B2 (en) * 2006-09-18 2017-05-09 John Nicholas and Kristin Gross Trust System and method of modifying ranking for internet accessible documents
US20080082578A1 (en) * 2006-09-29 2008-04-03 Andrew Hogue Displaying search results on a one or two dimensional graph
US9069750B2 (en) 2006-10-10 2015-06-30 Abbyy Infopoisk Llc Method and system for semantic searching of natural language texts
US9495358B2 (en) 2006-10-10 2016-11-15 Abbyy Infopoisk Llc Cross-language text clustering
US9098489B2 (en) 2006-10-10 2015-08-04 Abbyy Infopoisk Llc Method and system for semantic searching
US9892111B2 (en) 2006-10-10 2018-02-13 Abbyy Production Llc Method and device to estimate similarity between documents having multiple segments
US9189482B2 (en) 2012-10-10 2015-11-17 Abbyy Infopoisk Llc Similar document search
US9075864B2 (en) 2006-10-10 2015-07-07 Abbyy Infopoisk Llc Method and system for semantic searching using syntactic and semantic analysis
US9817902B2 (en) 2006-10-27 2017-11-14 Netseer Acquisition, Inc. Methods and apparatus for matching relevant content to user intention
US20080133365A1 (en) * 2006-11-21 2008-06-05 Benjamin Sprecher Targeted Marketing System
US8086600B2 (en) * 2006-12-07 2011-12-27 Google Inc. Interleaving search results
US7966321B2 (en) 2007-01-17 2011-06-21 Google Inc. Presentation of local results
US8326858B2 (en) * 2007-01-17 2012-12-04 Google Inc. Synchronization of fixed and mobile data
US8966407B2 (en) 2007-01-17 2015-02-24 Google Inc. Expandable homepage modules
US7966309B2 (en) 2007-01-17 2011-06-21 Google Inc. Providing relevance-ordered categories of information
US8005822B2 (en) 2007-01-17 2011-08-23 Google Inc. Location in search queries
US7975301B2 (en) * 2007-03-05 2011-07-05 Microsoft Corporation Neighborhood clustering for web spam detection
US8166021B1 (en) 2007-03-30 2012-04-24 Google Inc. Query phrasification
US8086594B1 (en) 2007-03-30 2011-12-27 Google Inc. Bifurcated document relevance scoring
US7693813B1 (en) 2007-03-30 2010-04-06 Google Inc. Index server architecture using tiered and sharded phrase posting lists
US7925655B1 (en) 2007-03-30 2011-04-12 Google Inc. Query scheduling using hierarchical tiers of index servers
US7702614B1 (en) 2007-03-30 2010-04-20 Google Inc. Index updating using segment swapping
US8166045B1 (en) 2007-03-30 2012-04-24 Google Inc. Phrase extraction using subphrase scoring
US8977631B2 (en) 2007-04-16 2015-03-10 Ebay Inc. Visualization of reputation ratings
US9535810B1 (en) 2007-04-24 2017-01-03 Wal-Mart Stores, Inc. Layout optimization
US7853589B2 (en) * 2007-04-30 2010-12-14 Microsoft Corporation Web spam page classification using query-dependent data
US7788254B2 (en) * 2007-05-04 2010-08-31 Microsoft Corporation Web page analysis using multiple graphs
US7908279B1 (en) 2007-05-25 2011-03-15 Amazon Technologies, Inc. Filtering invalid tokens from a document using high IDF token filtering
US7814107B1 (en) 2007-05-25 2010-10-12 Amazon Technologies, Inc. Generating similarity scores for matching non-identical data strings
US8046372B1 (en) 2007-05-25 2011-10-25 Amazon Technologies, Inc. Duplicate entry detection system and method
US20080301033A1 (en) * 2007-06-01 2008-12-04 Netseer, Inc. Method and apparatus for optimizing long term revenues in online auctions
US8117223B2 (en) * 2007-09-07 2012-02-14 Google Inc. Integrating external related phrase information into a phrase-based indexing information retrieval system
US20090112843A1 (en) * 2007-10-29 2009-04-30 International Business Machines Corporation System and method for providing differentiated service levels for search index
US20090119572A1 (en) * 2007-11-02 2009-05-07 Marja-Riitta Koivunen Systems and methods for finding information resources
US7895225B1 (en) * 2007-12-06 2011-02-22 Amazon Technologies, Inc. Identifying potential duplicates of a document in a document corpus
AR070182A1 (en) * 2008-01-15 2010-03-17 Thomson Reuters Glo Resources SYSTEMS METHODS AND SOFTWARE FOR THE REDACTION AND ANALYSIS OF MULTIFUNCTIONAL DOCUMENTS
US8752184B1 (en) 2008-01-17 2014-06-10 Google Inc. Spam detection for user-generated multimedia items based on keyword stuffing
US8745056B1 (en) 2008-03-31 2014-06-03 Google Inc. Spam detection for user-generated multimedia items based on concept clustering
WO2009111631A1 (en) * 2008-03-05 2009-09-11 Chacha Search, Inc. Method and system for triggering a search request
US8171020B1 (en) * 2008-03-31 2012-05-01 Google Inc. Spam detection for user-generated multimedia items based on appearance in popular queries
US8219385B2 (en) * 2008-04-08 2012-07-10 Incentive Targeting, Inc. Computer-implemented method and system for conducting a search of electronically stored information
US10387892B2 (en) * 2008-05-06 2019-08-20 Netseer, Inc. Discovering relevant concept and context for content node
US8788476B2 (en) * 2008-08-15 2014-07-22 Chacha Search, Inc. Method and system of triggering a search request
US8417695B2 (en) * 2008-10-30 2013-04-09 Netseer, Inc. Identifying related concepts of URLs and domain names
GB2472250A (en) * 2009-07-31 2011-02-02 Stephen Timothy Morris Method for determining document relevance
US8713078B2 (en) 2009-08-13 2014-04-29 Samsung Electronics Co., Ltd. Method for building taxonomy of topics and categorizing videos
US8838576B2 (en) * 2009-10-12 2014-09-16 Yahoo! Inc. Posting list intersection parallelism in query processing
US8386239B2 (en) 2010-01-25 2013-02-26 Holovisions LLC Multi-stage text morphing
US20110202521A1 (en) * 2010-01-28 2011-08-18 Jason Coleman Enhanced database search features and methods
US8392175B2 (en) 2010-02-01 2013-03-05 Stratify, Inc. Phrase-based document clustering with automatic phrase extraction
US8150859B2 (en) * 2010-02-05 2012-04-03 Microsoft Corporation Semantic table of contents for search results
US8161073B2 (en) 2010-05-05 2012-04-17 Holovisions, LLC Context-driven search
US8639773B2 (en) * 2010-06-17 2014-01-28 Microsoft Corporation Discrepancy detection for web crawling
US8655648B2 (en) * 2010-09-01 2014-02-18 Microsoft Corporation Identifying topically-related phrases in a browsing sequence
US9069842B2 (en) * 2010-09-28 2015-06-30 The Mitre Corporation Accessing documents using predictive word sequences
US9015043B2 (en) 2010-10-01 2015-04-21 Google Inc. Choosing recognized text from a background environment
US9424351B2 (en) 2010-11-22 2016-08-23 Microsoft Technology Licensing, Llc Hybrid-distribution model for search engine indexes
US9195745B2 (en) 2010-11-22 2015-11-24 Microsoft Technology Licensing, Llc Dynamic query master agent for query execution
US8478704B2 (en) 2010-11-22 2013-07-02 Microsoft Corporation Decomposable ranking for efficient precomputing that selects preliminary ranking features comprising static ranking features and dynamic atom-isolated components
US8620907B2 (en) * 2010-11-22 2013-12-31 Microsoft Corporation Matching funnel for large document index
US9529908B2 (en) 2010-11-22 2016-12-27 Microsoft Technology Licensing, Llc Tiering of posting lists in search engine index
US9342582B2 (en) 2010-11-22 2016-05-17 Microsoft Technology Licensing, Llc Selection of atoms for search engine retrieval
US9275001B1 (en) 2010-12-01 2016-03-01 Google Inc. Updating personal content streams based on feedback
US8626781B2 (en) * 2010-12-29 2014-01-07 Microsoft Corporation Priority hash index
US8880517B2 (en) 2011-02-18 2014-11-04 Microsoft Corporation Propagating signals across a web graph
US8332415B1 (en) * 2011-03-16 2012-12-11 Google Inc. Determining spam in information collected by a source
JP5492814B2 (en) * 2011-03-28 2014-05-14 デジタルア−ツ株式会社 SEARCH DEVICE, SEARCH SYSTEM, METHOD, AND PROGRAM
US8849835B1 (en) * 2011-05-10 2014-09-30 Google Inc. Reconciling data
US8782058B2 (en) * 2011-10-12 2014-07-15 Desire2Learn Incorporated Search index dictionary
US8909643B2 (en) * 2011-12-09 2014-12-09 International Business Machines Corporation Inferring emerging and evolving topics in streaming text
US8868536B1 (en) 2012-01-04 2014-10-21 Google Inc. Real time map spam detection
US8892422B1 (en) 2012-07-09 2014-11-18 Google Inc. Phrase identification in a sequence of words
US10311085B2 (en) 2012-08-31 2019-06-04 Netseer, Inc. Concept-level user intent profile extraction and applications
US9529907B2 (en) 2012-12-31 2016-12-27 Google Inc. Hold back and real time ranking of results in a streaming matching system
US9501506B1 (en) 2013-03-15 2016-11-22 Google Inc. Indexing system
US9483568B1 (en) 2013-06-05 2016-11-01 Google Inc. Indexing system
IN2013MU02217A (en) * 2013-07-01 2015-06-12 Tata Consultancy Services Ltd
US10565188B2 (en) * 2014-05-30 2020-02-18 Macy's West Stores, Inc. System and method for performing a pattern matching search
US10275485B2 (en) 2014-06-10 2019-04-30 Google Llc Retrieving context from previous sessions
US10347240B2 (en) * 2015-02-26 2019-07-09 Nantmobile, Llc Kernel-based verbal phrase splitting devices and methods
US10229219B2 (en) * 2015-05-01 2019-03-12 Facebook, Inc. Systems and methods for demotion of content items in a feed
US11392582B2 (en) * 2015-10-15 2022-07-19 Sumo Logic, Inc. Automatic partitioning
RU2618374C1 (en) 2015-11-05 2017-05-03 Общество с ограниченной ответственностью "Аби ИнфоПоиск" Identifying collocations in the texts in natural language
KR102025819B1 (en) * 2017-03-29 2019-09-26 중앙대학교 산학협력단 Device and method for constructing dynamic-terms identification system in user gererated contents
CN107391554B (en) * 2017-06-07 2021-10-01 中国人民解放军国防科学技术大学 Efficient distributed locality sensitive hashing method
JP2021092925A (en) * 2019-12-09 2021-06-17 株式会社東芝 Data generating device and data generating method
CN113312898B (en) * 2020-02-26 2024-03-01 深信服科技股份有限公司 Corpus processing method, corpus processing equipment, storage medium and corpus processing device
US11593439B1 (en) * 2022-05-23 2023-02-28 Onetrust Llc Identifying similar documents in a file repository using unique document signatures

Family Cites Families (160)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5321833A (en) 1990-08-29 1994-06-14 Gte Laboratories Incorporated Adaptive ranking system for information retrieval
US5278980A (en) * 1991-08-16 1994-01-11 Xerox Corporation Iterative technique for phrase query formation and an information retrieval system employing same
US5523946A (en) 1992-02-11 1996-06-04 Xerox Corporation Compact encoding of multi-lingual translation dictionaries
US5353401A (en) * 1992-11-06 1994-10-04 Ricoh Company, Ltd. Automatic interface layout generator for database systems
JPH0756933A (en) 1993-06-24 1995-03-03 Xerox Corp Method for retrieval of document
US5692176A (en) 1993-11-22 1997-11-25 Reed Elsevier Inc. Associative text search and retrieval system
US5715443A (en) * 1994-07-25 1998-02-03 Apple Computer, Inc. Method and apparatus for searching for information in a data processing system and for providing scheduled search reports in a summary format
JP3669016B2 (en) * 1994-09-30 2005-07-06 株式会社日立製作所 Document information classification device
US5694593A (en) * 1994-10-05 1997-12-02 Northeastern University Distributed computer database system and method
US5758257A (en) 1994-11-29 1998-05-26 Herz; Frederick System and method for scheduling broadcast of and access to video programs and other data using customer profiles
US6460036B1 (en) 1994-11-29 2002-10-01 Pinpoint Incorporated System and method for providing customized electronic newspapers and target advertisements
US5745602A (en) * 1995-05-01 1998-04-28 Xerox Corporation Automatic method of selecting multi-word key phrases from a document
US5659732A (en) * 1995-05-17 1997-08-19 Infoseek Corporation Document retrieval over networks wherein ranking and relevance scores are computed at the client for multiple database documents
US5724571A (en) 1995-07-07 1998-03-03 Sun Microsystems, Inc. Method and apparatus for generating query responses in a computer-based document retrieval system
US5668987A (en) * 1995-08-31 1997-09-16 Sybase, Inc. Database system with subquery optimizer
US6366933B1 (en) 1995-10-27 2002-04-02 At&T Corp. Method and apparatus for tracking and viewing changes on the web
US5757917A (en) * 1995-11-01 1998-05-26 First Virtual Holdings Incorporated Computerized payment system for purchasing goods and services on the internet
US6098034A (en) 1996-03-18 2000-08-01 Expert Ease Development, Ltd. Method for standardizing phrasing in a document
US7051024B2 (en) 1999-04-08 2006-05-23 Microsoft Corporation Document summarizer for word processors
US5924108A (en) 1996-03-29 1999-07-13 Microsoft Corporation Document summarizer for word processors
US5721897A (en) * 1996-04-09 1998-02-24 Rubinstein; Seymour I. Browse by prompted keyword phrases with an improved user interface
US5826261A (en) 1996-05-10 1998-10-20 Spencer; Graham System and method for querying multiple, distributed databases by selective sharing of local relative significance information for terms related to the query
US5915249A (en) 1996-06-14 1999-06-22 Excite, Inc. System and method for accelerated query evaluation of very large full-text databases
EP0822502A1 (en) 1996-07-31 1998-02-04 BRITISH TELECOMMUNICATIONS public limited company Data access system
US5920854A (en) 1996-08-14 1999-07-06 Infoseek Corporation Real-time document collection search engine with phrase indexing
US6085186A (en) 1996-09-20 2000-07-04 Netbot, Inc. Method and system using information written in a wrapper description language to execute query on a network
US20030093790A1 (en) * 2000-03-28 2003-05-15 Logan James D. Audio and video program recording, editing and playback systems using metadata
JP3584848B2 (en) * 1996-10-31 2004-11-04 富士ゼロックス株式会社 Document processing device, item search device, and item search method
US5960383A (en) 1997-02-25 1999-09-28 Digital Equipment Corporation Extraction of key sections from texts using automatic indexing techniques
US6055540A (en) 1997-06-13 2000-04-25 Sun Microsystems, Inc. Method and apparatus for creating a category hierarchy for classification of documents
US6470307B1 (en) 1997-06-23 2002-10-22 National Research Council Of Canada Method and apparatus for automatically identifying keywords within a document
US5983216A (en) * 1997-09-12 1999-11-09 Infoseek Corporation Performing automated document collection and selection by providing a meta-index with meta-index values indentifying corresponding document collections
US6018733A (en) * 1997-09-12 2000-01-25 Infoseek Corporation Methods for iteratively and interactively performing collection selection in full text searches
US5845278A (en) * 1997-09-12 1998-12-01 Inioseek Corporation Method for automatically selecting collections to search in full text searches
US5956722A (en) 1997-09-23 1999-09-21 At&T Corp. Method for effective indexing of partially dynamic documents
US6542888B2 (en) 1997-11-26 2003-04-01 International Business Machines Corporation Content filtering for electronic documents generated in multiple foreign languages
JP4183311B2 (en) 1997-12-22 2008-11-19 株式会社リコー Document annotation method, annotation device, and recording medium
US6185558B1 (en) 1998-03-03 2001-02-06 Amazon.Com, Inc. Identifying the items most relevant to a current query based on items selected in connection with similar queries
JP3664874B2 (en) 1998-03-28 2005-06-29 松下電器産業株式会社 Document search device
US6638314B1 (en) * 1998-06-26 2003-10-28 Microsoft Corporation Method of web crawling utilizing crawl numbers
US6363377B1 (en) 1998-07-30 2002-03-26 Sarnoff Corporation Search data processor
US6377949B1 (en) 1998-09-18 2002-04-23 Tacit Knowledge Systems, Inc. Method and apparatus for assigning a confidence level to a term within a user knowledge profile
US6366911B1 (en) * 1998-09-28 2002-04-02 International Business Machines Corporation Partitioning of sorted lists (containing duplicate entries) for multiprocessors sort and merge
US6415283B1 (en) 1998-10-13 2002-07-02 Orack Corporation Methods and apparatus for determining focal points of clusters in a tree structure
US7058589B1 (en) 1998-12-17 2006-06-06 Iex Corporation Method and system for employee work scheduling
US6862710B1 (en) 1999-03-23 2005-03-01 Insightful Corporation Internet navigation using soft hyperlinks
JP4021583B2 (en) 1999-04-08 2007-12-12 富士通株式会社 Information search apparatus, information search method, and recording medium storing program for realizing the method
US6430539B1 (en) * 1999-05-06 2002-08-06 Hnc Software Predictive modeling of consumer financial behavior
US7089236B1 (en) 1999-06-24 2006-08-08 Search 123.Com, Inc. Search engine interface
US6601026B2 (en) 1999-09-17 2003-07-29 Discern Communications, Inc. Information retrieval by natural language querying
US6996775B1 (en) 1999-10-29 2006-02-07 Verizon Laboratories Inc. Hypervideo: information retrieval using time-related multimedia:
US6684183B1 (en) 1999-12-06 2004-01-27 Comverse Ltd. Generic natural language service creation environment
US6963867B2 (en) 1999-12-08 2005-11-08 A9.Com, Inc. Search query processing to provide category-ranked presentation of search results
US6772150B1 (en) 1999-12-10 2004-08-03 Amazon.Com, Inc. Search query refinement using related search phrases
CA2293064C (en) 1999-12-22 2004-05-04 Ibm Canada Limited-Ibm Canada Limitee Method and apparatus for analyzing data retrieval using index scanning
US6981040B1 (en) 1999-12-28 2005-12-27 Utopy, Inc. Automatic, personalized online information and product services
US6820237B1 (en) 2000-01-21 2004-11-16 Amikanow! Corporation Apparatus and method for context-based highlighting of an electronic document
US6883135B1 (en) 2000-01-28 2005-04-19 Microsoft Corporation Proxy server using a statistical model
US6654739B1 (en) 2000-01-31 2003-11-25 International Business Machines Corporation Lightweight document clustering
US6571240B1 (en) 2000-02-02 2003-05-27 Chi Fai Ho Information processing for searching categorizing information in a document based on a categorization hierarchy and extracted phrases
US7137065B1 (en) 2000-02-24 2006-11-14 International Business Machines Corporation System and method for classifying electronically posted documents
US20060143714A1 (en) 2000-03-09 2006-06-29 Pkware, Inc. System and method for manipulating and managing computer archive files
US6859800B1 (en) 2000-04-26 2005-02-22 Global Information Research And Technologies Llc System for fulfilling an information need
AU2001261505A1 (en) 2000-05-11 2001-11-20 University Of Southern California Machine translation techniques
US6691106B1 (en) 2000-05-23 2004-02-10 Intel Corporation Profile driven instant web portal
WO2001098942A2 (en) 2000-06-19 2001-12-27 Lernout & Hauspie Speech Products N.V. Package driven parsing using structure function grammar
US20020078090A1 (en) 2000-06-30 2002-06-20 Hwang Chung Hee Ontological concept-based, user-centric text summarization
EP1182577A1 (en) 2000-08-18 2002-02-27 SER Systeme AG Produkte und Anwendungen der Datenverarbeitung Associative memory
KR100426382B1 (en) 2000-08-23 2004-04-08 학교법인 김포대학 Method for re-adjusting ranking document based cluster depending on entropy information and Bayesian SOM(Self Organizing feature Map)
US7017114B2 (en) 2000-09-20 2006-03-21 International Business Machines Corporation Automatic correlation method for generating summaries for text documents
US20020147578A1 (en) 2000-09-29 2002-10-10 Lingomotors, Inc. Method and system for query reformulation for searching of information
US20020065857A1 (en) 2000-10-04 2002-05-30 Zbigniew Michalewicz System and method for analysis and clustering of documents for search engine
CA2322599A1 (en) * 2000-10-06 2002-04-06 Ibm Canada Limited-Ibm Canada Limitee System and method for workflow control of contractual activities
JP2002132789A (en) 2000-10-19 2002-05-10 Hitachi Ltd Document retrieving method
JP2002169834A (en) * 2000-11-20 2002-06-14 Hewlett Packard Co <Hp> Computer and method for making vector analysis of document
US20020091671A1 (en) 2000-11-23 2002-07-11 Andreas Prokoph Method and system for data retrieval in large collections of data
JP2002207760A (en) 2001-01-10 2002-07-26 Hitachi Ltd Document retrieval method, executing device thereof, and storage medium with its processing program stored therein
US6778980B1 (en) 2001-02-22 2004-08-17 Drugstore.Com Techniques for improved searching of electronically stored information
US6741984B2 (en) 2001-02-23 2004-05-25 General Electric Company Method, system and storage medium for arranging a database
US6721728B2 (en) 2001-03-02 2004-04-13 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration System, method and apparatus for discovering phrases in a database
US6741981B2 (en) * 2001-03-02 2004-05-25 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration (Nasa) System, method and apparatus for conducting a phrase search
US6697793B2 (en) 2001-03-02 2004-02-24 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration System, method and apparatus for generating phrases from a database
US6823333B2 (en) 2001-03-02 2004-11-23 The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration System, method and apparatus for conducting a keyterm search
US7194483B1 (en) 2001-05-07 2007-03-20 Intelligenxia, Inc. Method, system, and computer program product for concept-based multi-dimensional analysis of unstructured information
US7171619B1 (en) 2001-07-05 2007-01-30 Sun Microsystems, Inc. Methods and apparatus for accessing document content
US6769016B2 (en) * 2001-07-26 2004-07-27 Networks Associates Technology, Inc. Intelligent SPAM detection system using an updateable neural analysis engine
US20030031996A1 (en) * 2001-08-08 2003-02-13 Adam Robinson Method and system for evaluating documents
US20030130993A1 (en) * 2001-08-08 2003-07-10 Quiver, Inc. Document categorization engine
US6778979B2 (en) 2001-08-13 2004-08-17 Xerox Corporation System for automatically generating queries
US6978274B1 (en) * 2001-08-31 2005-12-20 Attenex Corporation System and method for dynamically evaluating latent concepts in unstructured documents
JP2003242176A (en) * 2001-12-13 2003-08-29 Sony Corp Information processing device and method, recording medium and program
US6741982B2 (en) 2001-12-19 2004-05-25 Cognos Incorporated System and method for retrieving data from a database system
US7356527B2 (en) 2001-12-19 2008-04-08 International Business Machines Corporation Lossy index compression
US7137062B2 (en) 2001-12-28 2006-11-14 International Business Machines Corporation System and method for hierarchical segmentation with latent semantic indexing in scale space
US7243092B2 (en) 2001-12-28 2007-07-10 Sap Ag Taxonomy generation for electronic documents
US7139756B2 (en) 2002-01-22 2006-11-21 International Business Machines Corporation System and method for detecting duplicate and similar documents
US7028045B2 (en) 2002-01-25 2006-04-11 International Business Machines Corporation Compressing index files in information retrieval
US7421660B2 (en) 2003-02-04 2008-09-02 Cataphora, Inc. Method and apparatus to visually present discussions for data mining purposes
JP4092933B2 (en) 2002-03-20 2008-05-28 富士ゼロックス株式会社 Document information retrieval apparatus and document information retrieval program
US7743045B2 (en) * 2005-08-10 2010-06-22 Google Inc. Detecting spam related and biased contexts for programmable search engines
US20030195937A1 (en) * 2002-04-16 2003-10-16 Kontact Software Inc. Intelligent message screening
US6877001B2 (en) * 2002-04-25 2005-04-05 Mitsubishi Electric Research Laboratories, Inc. Method and system for retrieving documents with spoken queries
NZ518744A (en) 2002-05-03 2004-08-27 Hyperbolex Ltd Electronic document indexing using word use nodes, node objects and link objects
US7085771B2 (en) * 2002-05-17 2006-08-01 Verity, Inc System and method for automatically discovering a hierarchy of concepts from a corpus of documents
US7028026B1 (en) 2002-05-28 2006-04-11 Ask Jeeves, Inc. Relevancy-based database retrieval and display techniques
CN1464413A (en) * 2002-06-13 2003-12-31 英业达股份有限公司 Overture data retrieving management system and method thereof based on web pages
JP4452012B2 (en) 2002-07-04 2010-04-21 ヒューレット・パッカード・カンパニー Document uniqueness evaluation method
JP2004046438A (en) 2002-07-10 2004-02-12 Nippon Telegr & Teleph Corp <Ntt> Text retrieval method and device, text retrieval program and storage medium storing text retrieval program
US20040034633A1 (en) 2002-08-05 2004-02-19 Rickard John Terrell Data search system and method using mutual subsethood measures
DE50202326D1 (en) * 2002-08-23 2005-03-31 Festo Ag & Co Kg sealing ring
US7151864B2 (en) 2002-09-18 2006-12-19 Hewlett-Packard Development Company, L.P. Information research initiated from a scanned image media
US7158983B2 (en) 2002-09-23 2007-01-02 Battelle Memorial Institute Text analysis technique
US6886010B2 (en) 2002-09-30 2005-04-26 The United States Of America As Represented By The Secretary Of The Navy Method for data and text mining and literature-based discovery
JP2004139150A (en) 2002-10-15 2004-05-13 Ricoh Co Ltd Document search system, program, and storage medium
US7970832B2 (en) * 2002-11-20 2011-06-28 Return Path, Inc. Electronic message delivery with estimation approaches and complaint, bond, and statistics panels
US20040133560A1 (en) 2003-01-07 2004-07-08 Simske Steven J. Methods and systems for organizing electronic documents
US7725544B2 (en) * 2003-01-24 2010-05-25 Aol Inc. Group based spam classification
GB2399427A (en) * 2003-03-12 2004-09-15 Canon Kk Apparatus for and method of summarising text
US7945567B2 (en) 2003-03-17 2011-05-17 Hewlett-Packard Development Company, L.P. Storing and/or retrieving a document within a knowledge base or document repository
US6947930B2 (en) 2003-03-21 2005-09-20 Overture Services, Inc. Systems and methods for interactive search query refinement
US7051023B2 (en) * 2003-04-04 2006-05-23 Yahoo! Inc. Systems and methods for generating concept units from search queries
US7149748B1 (en) 2003-05-06 2006-12-12 Sap Ag Expanded inverted index
US7664732B2 (en) * 2003-05-31 2010-02-16 Nhn Corporation Method of managing websites registered in search engine and a system thereof
US7272853B2 (en) * 2003-06-04 2007-09-18 Microsoft Corporation Origination/destination features and lists for spam prevention
US7051014B2 (en) 2003-06-18 2006-05-23 Microsoft Corporation Utilizing information redundancy to improve text searches
US7254580B1 (en) 2003-07-31 2007-08-07 Google Inc. System and method for selectively searching partitions of a database
JP2005056233A (en) * 2003-08-06 2005-03-03 Nec Corp Mobile communication device, and method and program for operating it for reception of email
US20050043940A1 (en) 2003-08-20 2005-02-24 Marvin Elder Preparing a data source for a natural language query
US20050060295A1 (en) * 2003-09-12 2005-03-17 Sensory Networks, Inc. Statistical classification of high-speed network data through content inspection
US20050071310A1 (en) 2003-09-30 2005-03-31 Nadav Eiron System, method, and computer program product for identifying multi-page documents in hypertext collections
US20050071328A1 (en) 2003-09-30 2005-03-31 Lawrence Stephen R. Personalization of web search
US7346839B2 (en) * 2003-09-30 2008-03-18 Google Inc. Information retrieval based on historical data
US7257564B2 (en) * 2003-10-03 2007-08-14 Tumbleweed Communications Corp. Dynamic message filtering
US7240064B2 (en) 2003-11-10 2007-07-03 Overture Services, Inc. Search engine with hierarchically stored indices
US20050160107A1 (en) 2003-12-29 2005-07-21 Ping Liang Advanced search, file system, and intelligent assistant agent
US7206389B1 (en) 2004-01-07 2007-04-17 Nuance Communications, Inc. Method and apparatus for generating a speech-recognition-based call-routing system
US20060294124A1 (en) 2004-01-12 2006-12-28 Junghoo Cho Unbiased page ranking
US20050198559A1 (en) 2004-03-08 2005-09-08 Kabushiki Kaisha Toshiba Document information management system, document information management program, and document information management method
US20050216564A1 (en) 2004-03-11 2005-09-29 Myers Gregory K Method and apparatus for analysis of electronic communications containing imagery
US20050256848A1 (en) 2004-05-13 2005-11-17 International Business Machines Corporation System and method for user rank search
US7155243B2 (en) 2004-06-15 2006-12-26 Tekelec Methods, systems, and computer program products for content-based screening of messaging service messages
JP2006026844A (en) * 2004-07-20 2006-02-02 Fujitsu Ltd Polishing pad, polishing device provided with it and sticking device
US7426507B1 (en) 2004-07-26 2008-09-16 Google, Inc. Automatic taxonomy generation in search results using phrases
US7599914B2 (en) 2004-07-26 2009-10-06 Google Inc. Phrase-based searching in an information retrieval system
US7711679B2 (en) 2004-07-26 2010-05-04 Google Inc. Phrase-based detection of duplicate documents in an information retrieval system
US7580921B2 (en) * 2004-07-26 2009-08-25 Google Inc. Phrase identification in an information retrieval system
US7702618B1 (en) 2004-07-26 2010-04-20 Google Inc. Information retrieval system for archiving multiple document versions
US7567959B2 (en) 2004-07-26 2009-07-28 Google Inc. Multiple index based information retrieval system
US7580929B2 (en) 2004-07-26 2009-08-25 Google Inc. Phrase-based personalization of searches in an information retrieval system
US7536408B2 (en) 2004-07-26 2009-05-19 Google Inc. Phrase-based indexing in an information retrieval system
US7584175B2 (en) 2004-07-26 2009-09-01 Google Inc. Phrase-based generation of document descriptions
US8407239B2 (en) * 2004-08-13 2013-03-26 Google Inc. Multi-stage query processing system and method for use with tokenspace repository
US8504565B2 (en) 2004-09-09 2013-08-06 William M. Pitts Full text search capabilities integrated into distributed file systems— incrementally indexing files
US7962510B2 (en) * 2005-02-11 2011-06-14 Microsoft Corporation Using content analysis to detect spam web pages
US20060200464A1 (en) 2005-03-03 2006-09-07 Microsoft Corporation Method and system for generating a document summary
US7552230B2 (en) * 2005-06-15 2009-06-23 International Business Machines Corporation Method and apparatus for reducing spam on peer-to-peer networks
US20080005064A1 (en) * 2005-06-28 2008-01-03 Yahoo! Inc. Apparatus and method for content annotation and conditional annotation retrieval in a search context
US7454449B2 (en) 2005-12-20 2008-11-18 International Business Machines Corporation Method for reorganizing a set of database partitions
WO2007123919A2 (en) 2006-04-18 2007-11-01 Gemini Design Technology, Inc. Method for ranking webpages via circuit simulation
US8117223B2 (en) 2007-09-07 2012-02-14 Google Inc. Integrating external related phrase information into a phrase-based indexing information retrieval system

Also Published As

Publication number Publication date
KR101190230B1 (en) 2012-10-12
EP1622053A1 (en) 2006-02-01
US7603345B2 (en) 2009-10-13
BRPI0503781A (en) 2006-03-14
JP4976666B2 (en) 2012-07-18
US8078629B2 (en) 2011-12-13
NO20053638D0 (en) 2005-07-26
US20110131223A1 (en) 2011-06-02
US20060294155A1 (en) 2006-12-28
ATE500559T1 (en) 2011-03-15
US20060018551A1 (en) 2006-01-26
NO20053638L (en) 2006-01-27
DE602005026609D1 (en) 2011-04-14
EP1622053B1 (en) 2011-03-02
CA2513850C (en) 2014-01-14
AU2005203240B2 (en) 2009-01-22
KR20060048779A (en) 2006-05-18
JP2006048683A (en) 2006-02-16
CN1728142B (en) 2011-01-12
AU2005203240A1 (en) 2006-02-09
CN1728142A (en) 2006-02-01
US7580921B2 (en) 2009-08-25

Similar Documents

Publication Publication Date Title
CA2513850A1 (en) Phrase identification in an information retrieval system
US10671676B2 (en) Multiple index based information retrieval system
US9990421B2 (en) Phrase-based searching in an information retrieval system
EP1622055B1 (en) Phrase-based indexing in an information retrieval system
CA2813644C (en) Phrase-based searching in an information retrieval system
CA2513851C (en) Phrase-based generation of document descriptions
US7426507B1 (en) Automatic taxonomy generation in search results using phrases
US7580929B2 (en) Phrase-based personalization of searches in an information retrieval system
US7702618B1 (en) Information retrieval system for archiving multiple document versions

Legal Events

Date Code Title Description
EEER Examination request
MKLA Lapsed

Effective date: 20170726