Suche Bilder Maps Play YouTube News Gmail Drive Mehr »
Anmelden
Nutzer von Screenreadern: Klicke auf diesen Link, um die Bedienungshilfen zu aktivieren. Dieser Modus bietet die gleichen Grundfunktionen, funktioniert aber besser mit deinem Reader.

Patentsuche

  1. Erweiterte Patentsuche
VeröffentlichungsnummerUS20140180692 A1
PublikationstypAnmeldung
AnmeldenummerUS 14/184,379
Veröffentlichungsdatum26. Juni 2014
Eingetragen19. Febr. 2014
Prioritätsdatum28. Febr. 2011
Auch veröffentlicht unterUS8688453
Veröffentlichungsnummer14184379, 184379, US 2014/0180692 A1, US 2014/180692 A1, US 20140180692 A1, US 20140180692A1, US 2014180692 A1, US 2014180692A1, US-A1-20140180692, US-A1-2014180692, US2014/0180692A1, US2014/180692A1, US20140180692 A1, US20140180692A1, US2014180692 A1, US2014180692A1
ErfinderSachindra Joshi, Shantanu Godbole
Ursprünglich BevollmächtigterNuance Communications, Inc.
Zitat exportierenBiBTeX, EndNote, RefMan
Externe Links: USPTO, USPTO-Zuordnung, Espacenet
Intent mining via analysis of utterances
US 20140180692 A1
Zusammenfassung
According to example configurations, a speech processing system can include a syntactic parser, a word extractor, word extraction rules, and an analyzer. The syntactic parser of the speech processing system parses the utterance to identify syntactic relationships amongst words in the utterance. The word extractor utilizes word extraction rules to identify groupings of related words in the utterance that most likely represent an intended meaning of the utterance. The analyzer in the speech processing system maps each set of the sets of words produced by the word extractor to a respective candidate intent value to produce a list of candidate intent values for the utterance. The analyzer is configured to select, from the list of candidate intent values (i.e., possible intended meanings) of the utterance, a particular candidate intent value as being representative of the intent (i.e., intended meaning) of the utterance.
Bilder(17)
Previous page
Next page
Ansprüche(2)
1. A method comprising:
parsing an utterance to identify syntactic relationships amongst words in the utterance;
creating sets of words from the utterance based on the syntactic relationships;
mapping each set of the sets of words to a respective candidate intent value to produce a list of candidate intent values for the utterance; and
selecting, from the list, a candidate intent value as being representative of an intent of the utterance.
2-23. (canceled)
Beschreibung
    BACKGROUND
  • [0001]
    Recent developments in computers and corresponding speech recognition software algorithms have made it possible to control computer equipment via spoken input. Thus, it is now becoming more common that users are able to control their computers, electronics, personal devices, call routing, etc., via speech input.
  • [0002]
    Speech recognition systems are highly complex and operate by matching an acoustic signature of an utterance with acoustic signatures of words in a language model. As an example, according to conventional speech recognition systems, a microphone first converts a received acoustic signature of an uttered word into an electrical signal. An A/D (analog-to-digital) converter is typically used to convert the electrical signal into a digital representation of the uttered word. A digital signal processor converts the captured electrical signal from the time domain to the frequency domain.
  • [0003]
    Generally, as another part of the speech recognition process, the digital signal processor breaks down the utterance into its spectral components. Typically, the amplitude or intensity of the digital signal at various frequencies and temporal locations are then compared to a language model to determine the one or more words that were uttered.
  • [0004]
    Certain conventional speech recognition systems can be used for classifying utterances. For example, a conventional speech recognition system can receive and convert an utterance into respective text. In certain instances, a conventional speech recognition system can be configured to classify the utterance based on a key word in the utterance.
  • BRIEF DESCRIPTION
  • [0005]
    Conventional speech recognition systems can suffer from drawbacks. For example, conventional speech recognition typically analyze an utterance for presence of words that indicate a likely class to which the utterance belongs. However, the accuracy of the classification can be quite low because an utterance may include many words, making the intended meaning of the utterance difficult to determine.
  • [0006]
    Embodiments herein deviate with respect to conventional speech recognition systems to provide accurate speech utterance classification. For example, one embodiment herein includes a speech processing system to identify one or more intended meanings of a received utterance based on word groupings derived from words detected in the utterance. In one embodiment, the speech processing system tags the utterance with respective one or more tags indicative of one or more classes believed to represent an intended meaning of the utterance.
  • [0007]
    More specifically, a speech processing system according to embodiments herein can include a syntactic parser, a word extractor, word pattern rules, and an analyzer. To identify an intended general meaning of the utterance, the speech processing system performs a number of tasks on a received utterance.
  • [0008]
    In one embodiment, the syntactic parser of the speech processing system parses a received utterance to identify syntactic relationships amongst words in the utterance. The word extractor creates sets of words using words in the utterance based at least in part on the syntactic relationships identified by the parser. As an example, the word extractor can utilize the word pattern rules to identify groupings of related words in the utterance that most likely represent an intended meaning of the utterance.
  • [0009]
    In one embodiment, the pattern rules specify which type and/or location of related words in the utterance that are to be used to create the sets of words. Accordingly, embodiments herein can include creating the sets of words can include utilizing the identified syntactic relationships of words to identify groupings of related words in the utterance; and applying a set of pattern rules or word extraction rules to the identified syntactic relationships to identify types and/or locations of words in the utterance to create the sets of words.
  • [0010]
    Subsequent to creating the groupings of words, an analyzer in the speech processing system maps each set of the sets of words produced by the word extractor to a respective candidate intent value to produce a list of candidate intent values for the utterance. Thus, if there are multiple candidate word groupings derived form the utterance, the received utterance can be mapped multiple candidate intent values. As its name suggests, a candidate intent value is a possible intended meaning of the utterance.
  • [0011]
    An utterance can have one or more intended meanings; the speech processing system can be configured to identify one or more most likely intended meanings of the utterance under test. In one embodiment, the analyzer is configured to select, from the list of possible intended meanings of the utterance, a particular candidate intent value as being representative of the intent (i.e., intended meaning) of the utterance.
  • [0012]
    The speech processing system can be configured to maintain statistical information for a pool of previously received utterances to determine the meaning of future utterances. For example, as previously discussed, the statistical information can indicate a frequency of receiving utterances of different intent types. The statistical information can be updated over time to reflect that a most recently received and analyzed utterance was assigned the particular candidate intent value as discussed above.
  • [0013]
    Selection of particular intent value amongst a group of possible intents for an utterance can be based on selection criteria employed by the analyzer. For example, in one embodiment, the analyzer identifies a frequency of occurrence that utterances in a pool of previously received utterances were of a same intent type as that of a first candidate intent value for the newly received utterance; the analyzer also identifies a frequency of occurrence that utterances in the pool of previously received utterances were of a same intent type as that of the second candidate intent value for the newly received utterance; and so on. As previously discussed, the analyzer can identify multiple possible classes in which to categorizes or classify the received utterance. In one embodiment, the analyzer then selects the a particular candidate intent value for assigning to the utterance depending on which of the possible candidate intent values (e.g., first candidate intent value, second candidate intent value, etc.) occurred more often in the pool for the previously received utterances. Thus, according to one embodiment, the analyzer can perform a frequency analysis and then sort the candidate meanings. As previously discussed, the selected candidate value indicates a likely dominant subject matter or theme of the utterance.
  • [0014]
    In yet further embodiments, the speech processing system includes a set of tags or labels. In one embodiment, a tagging resource (such as the analyzer or other suitable resource) identifies an appropriate tag that is representative of the intent value (or selected meaning) selected for the utterance. The tagging resource then tags the utterance with the appropriate tag to indicate a likely dominant subject matter intended by words in the received utterance. Accordingly, embodiments herein can include classifying one or more received utterances using tags.
  • [0015]
    As discussed above and below in further embodiments, techniques herein are well suited for use in software and/or hardware applications implementing speech recognition and classification of utterances based on intended meanings. However, it should be noted that embodiments herein are not limited to use in such applications and that the techniques discussed herein are well suited for other applications as well.
  • [0016]
    These and other embodiments are discussed in more detail below.
  • [0017]
    As mentioned above, note that embodiments herein can include a configuration of one or more computerized devices, workstations, handheld or laptop computers, or the like to carry out and/or support any or all of the method operations disclosed herein. In other words, one or more computerized devices or processors can be programmed and/or configured to operate as explained herein to carry out different embodiments of the invention.
  • [0018]
    Yet other embodiments herein include software programs to perform the steps and operations summarized above and disclosed in detail below. One such embodiment comprises a computer program product including a non-transitory computer-readable storage medium on which software instructions are encoded for subsequent execution. The instructions, when executed in a computerized device having a processor, program and/or cause the processor to perform the operations disclosed herein. Such arrangements are typically provided as software, code, instructions, and/or other data (e.g., data structures) arranged or encoded on a non-transitory computer readable storage medium such as an optical medium (e.g., CD-ROM), floppy disk, hard disk, memory stick, etc., or other a medium such as firmware or microcode in one or more ROM, RAM, PROM, etc., or as an Application Specific Integrated Circuit (ASIC), etc. The software or firmware or other such configurations can be installed onto a computerized device to cause the computerized device to perform the techniques explained herein.
  • [0019]
    Accordingly, one particular embodiment of the present disclosure is directed to a computer program product that includes a computer readable storage medium having instructions stored thereon for speech recognition such as converting of an utterance to corresponding text. For example, in one embodiment, the instructions, when executed by a processor of a respective computer device, cause the processor to: parse an utterance to identify syntactic relationships amongst words in the utterance; create or group sets of words from the utterance based at least in part on the syntactic relationships; map each set of the sets of words to a respective candidate intent value; produce a list of candidate intent values for the utterance based on the mapping; and select, from the list, a candidate intent value as being representative of an intent of the utterance.
  • [0020]
    The ordering of the steps has been added for clarity sake. These steps can be performed in any suitable order.
  • [0021]
    Other embodiments of the present disclosure include software programs and/or respective hardware to perform any of the method embodiment steps and operations summarized above and disclosed in detail below.
  • [0022]
    It is to be understood that the system, method, apparatus, instructions on computer readable storage media, etc., as discussed herein can be embodied strictly as a software program, as a hybrid of software and hardware, or as hardware alone such as within a processor, or within an operating system or a within a software application. Example embodiments of the invention may be implemented within products and/or software applications such as those manufactured by Nuance Communications, Inc., Burlington, Mass., USA.
  • [0023]
    Additionally, although each of the different features, techniques, configurations, etc., herein may be discussed in different places of this disclosure, it is intended that each of the concepts can be executed independently of each other or, where suitable, the concepts can be used in combination with each other. Accordingly, the one or more present inventions as described herein can be embodied and viewed in many different ways.
  • [0024]
    Also, note that this preliminary discussion of embodiments herein does not specify every embodiment and/or incrementally novel aspect of the present disclosure or claimed invention(s). Instead, this brief description only presents general embodiments and corresponding points of novelty over conventional techniques. For additional details and/or possible perspectives (permutations) of the invention(s), and additional points of novelty, the reader is directed to the Detailed Description section and corresponding figures of the present disclosure as further discussed below.
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • [0025]
    FIG. 1 is an example block diagram of a speech processing system to perform intent mining for one or more utterances according to embodiments herein.
  • [0026]
    FIG. 2 is an example diagram illustrating a sample utterance and corresponding relationship information generated by a syntactic parser according to embodiments herein.
  • [0027]
    FIG. 3 is an example diagram illustrating a sample utterance and corresponding relationship information generated by a syntactic parser according to embodiments herein.
  • [0028]
    FIG. 4 is an example diagram illustrating pattern or word extraction rules according to embodiments herein.
  • [0029]
    FIG. 5 is an example diagram illustrating a listing of different utterances that have been classified under a representative intent value according to embodiments herein.
  • [0030]
    FIG. 6 is an example diagram illustrating a listing of different utterances that have been classified under a representative intent value according to embodiments herein.
  • [0031]
    FIG. 7 is an example diagram illustrating a listing of different utterances that have been classified under a representative intent value according to embodiments herein.
  • [0032]
    FIG. 8 is an example diagram illustrating a listing of different utterances that have been classified under a representative intent value according to embodiments herein.
  • [0033]
    FIG. 9 is an example diagram illustrating sets of words extracted from utterances and corresponding frequency information of occurrence according to embodiments herein.
  • [0034]
    FIG. 10 is an example diagram illustrating application of intent mining to an example according to embodiments herein.
  • [0035]
    FIG. 11 is an example diagram illustrating intent values (e.g., representative word groupings) assigned to a respective subject matter label according to embodiments herein.
  • [0036]
    FIG. 12 is an example diagram illustrating intent values and corresponding assigned subject matter labels according to embodiments herein.
  • [0037]
    FIG. 13 is a diagram illustrating an example computer architecture for executing a speech processing system according to embodiments herein.
  • [0038]
    FIG. 14 is a flowchart illustrating an example method of implementing a speech processing system according to embodiments herein.
  • [0039]
    FIGS. 15 and 16 combine to form a flowchart illustrating an example method of implementing speech processing system according to embodiments herein.
  • [0040]
    The foregoing and other objects, features, and advantages of the invention will be apparent from the following more particular description of preferred embodiments herein, as illustrated in the accompanying drawings in which like reference characters refer to the same parts throughout the different views. The drawings are not necessarily to scale, with emphasis instead being placed upon illustrating the embodiments, principles, concepts, etc.
  • DETAILED DESCRIPTION
  • [0041]
    According to one embodiment, a speech processing system includes a syntactic parser, a word extractor, word extraction rules, and an analyzer. The syntactic parser of the speech processing system parses the utterance to identify syntactic relationships amongst words in the utterance. In accordance with the syntactic relationship information, the word extractor utilizes the word extraction rules to identify groupings of related words in the utterance that most likely represent an intended meaning of the utterance. The analyzer in the speech processing system maps each set of the sets of words produced by the word extractor to a respective candidate intent value to produce a list of candidate intent values for the utterance. The candidate intent values represent a possible intended meaning of the utterance. The analyzer is configured to select, from the list of candidate intent values (i.e., possible intended meanings) of the utterance, a particular candidate intent value as being representative of a dominant intent of the utterance.
  • [0042]
    Now, referring to the figures, FIG. 1 is an example diagram illustrating a speech processing system 100 according to embodiments herein. As shown, speech processing system 100 includes a syntactic parser 115, word extractor 140, and an analyzer 150. The speech processing system 100 can receive one or more utterances 110-1, 110-2, 100-3, . . . (collectively, utterances 110) in response to a query such as “How may I assist you?”
  • [0043]
    As previously discussed, embodiments herein deviate with respect to conventional speech recognition systems to provide accurate speech utterance classification. For example, one embodiment herein includes a speech processing system 100 to identify one or more intended meanings of a received utterance 110. The speech processing system 100 can tag the utterance with respective one or more tags indicative of one or more classes that appear to best represent an intended meaning of the utterance 110.
  • [0044]
    To identify an intended general meaning of a received utterance or sequence of inputted text, the speech processing system 110 performs a number of tasks on a received utterance. For example, the syntactic parser 115 of the speech processing system 100 converts the utterance to respective text and parses the text in the utterance 110 to produce syntactic relationship information 120. The syntactic relationship information 120 indicates syntactic relationships amongst the text-based words in the utterance 110.
  • [0045]
    The word extractor 140 creates sets of words (e.g., candidate word groupings 145) using words in the utterance 110 based at least in part on the syntactic relationship information 120 identified by the syntactic parser 115. As an example, the word extractor 140 can utilize the word extraction rules 125 to identify one or more groupings of related words in the utterance that most likely represent an intended meaning of the utterances.
  • [0046]
    In one embodiment, the word extraction rules 125 includes patterns and/or templates that indicate, for example, which types of words, locations of words, etc., in the utterance will be used to create respective word groupings.
  • [0047]
    In accordance with further embodiments, the word extraction rules 125 specify which type and/or location of related words in the utterances are to be used to create the sets of words for a respective utterance. Accordingly, embodiments herein include creating the sets of words for a respective utterance under test can include utilizing the identified syntactic relationship information 120 to identify groupings of related words in the respective utterance; and applying a set of word extraction rules 125 to the identified syntactic relationships and utterance under test to identify types and/or locations of words in the utterance to create the sets of words.
  • [0048]
    Subsequent to creating one or more candidate word groupings 145 for a respective utterance, an analyzer 150 in the speech processing system 150 maps each set of the sets of words (e.g., candidate word groupings 145) produced by the word extractor 140 to a respective candidate intent value to produce a list of candidate intent values for the utterance. As its name suggests, a candidate intent value is a possible intended meaning of the utterance.
  • [0049]
    Note that a candidate word grouping produced by the speech processing system 100 for a respective utterance can include any number of words such as a single word or two or more words. The word extractor 140 can produce one or multiple candidate word groupings 145 for each utterance.
  • [0050]
    An utterance can have one or more intended meanings; the speech processing system 100 can be configured to identify one or more most likely intended meanings of an utterance under test. If there is only one word grouping and corresponding candidate intent value produced for an utterance under test, then the analyzer can assign the single intended meaning of the single candidate intent value to the utterance.
  • [0051]
    If analysis of an utterance produces multiple possible candidates, subsequent to determining the list of possible intended meanings (e.g., candidate intent values) of an utterance under test, the analyzer 150 applies further processing to narrow down the multiple possible intended meanings of the utterance to a at least most likely representative of the utterance. For example, in one embodiment, the analyzer is configured to select, from the list of possible intended meanings of the utterance as indicated by the candidate word groupings 145, a particular candidate intent value as being representative of the intent (i.e., intended meaning) of the utterance.
  • [0052]
    This technique of processing can be applied to each of the received utterances 110. Thus, in other words, the speech processing system 100 can be configured to classify the received utterances based on intended meaning.
  • [0053]
    The speech processing system 10 can be configured to maintain statistical information 160 for a pool of previously received utterances. For example, as previously discussed, the statistical information 160 can indicate an historical frequency of receiving utterances of different intent types. For newly received utterances and determination of respective intended meanings, the statistical information 160 can be updated over time to reflect that a most recently received and analyzed utterance was assigned a particular candidate intent value. Thus, over time, the statistical information 160 can be configured to keep track of how often the speech processing system 100 receives utterances of each particular intent type.
  • [0054]
    Selection of an intent value amongst a group of possible intents to assign to a newly received utterance under test can be based on selection criteria employed by the analyzer 150. In such an embodiment, the analyzer 150 selects the a particular candidate intent value for assigning to the utterance depending on which of the possible candidate intent values (e.g., first candidate intent value, second candidate intent value, etc.) occurred more often in the pool for the previously received utterances as specified by the statistical information 160. Thus, according to one embodiment, the analyzer 150 can perform a frequency analysis and then sort the candidate meanings to perform intent mining. As previously discussed, the selected candidate value indicates a likely dominant subject matter or theme of the utterance.
  • [0055]
    One embodiment includes so-called clustering. In such an embodiment, for each intent, the analyzer 150 compute such as respective frequency and sorts the intents. For (i=11, 12, . . . , i_n), let U={u1, u2, . . . , u_m} be the utterance containing i, For j in U, if the j is already not covered, then j is covered by i. Mark j covered.
  • [0056]
    Note that any suitable method can be implemented to perform clustering.
  • [0057]
    In yet further embodiments, the speech processing system 100 can include a set of tags for assigning to the utterances. In one embodiment, a tagging resource (such as the analyzer 150 or other suitable resource) identifies an appropriate tag that is representative of the intent value (or selected meaning) for an utterance. The tagging resource then tags the utterance with the appropriate tag to indicate a likely dominant subject matter intended by words in the received utterance. Accordingly, embodiments herein can include classifying one or more received utterances and then applying tags to identify a respective intended meaning.
  • [0058]
    FIG. 2 is an example diagram illustrating a sample utterance and corresponding relationship information 120 according to embodiments herein. Syntactic parser 115 receives the example utterance 110-1 “I would like to speak with a customer representative.” Based on parsing, the syntactic parser 115 produces syntactic relationship information 120-1 (p=pronoun, voltage=verb, a=adjective, network=noun, . . . ) as shown. By way of a non-limiting example, a Link Grammar parser can be used to parse the received utterance 110-1.
  • [0059]
    FIG. 3 is an example diagram illustrating a sample utterance and corresponding relationship information according to embodiments herein. Syntactic parser 115 receives the example utterance “I would like to speak with a customer service representative.” Based on parsing rules, the syntactic parser 115 produces syntactic relationship information 120-2 as shown (e.g., p=pronoun, voltage=verb, a=adjective, network=noun, . . . ).
  • [0060]
    FIG. 4 is an example diagram illustrating word extraction rules 125 according to embodiments herein. In one embodiment, in conjunction with the syntactic relationship information 125, the word extraction rules 125 are used to identify the location of different types of words in a received utterance that are to be used to generate respective candidate word groupings 145.
  • [0061]
    During operation, the word extractor 140 applies these rules to produce candidate word groupings 145. In the above example FIG. 2, the word extractor 140 applies word extraction rule 125-1 (FIG. 4) to example utterance 110-1 “I need to change my next order.” For this utterance 110-1 (FIG. 2) and using word extraction rule 125-1, Lword=“change” and Rword=“order” to produce the candidate word grouping change_order based on format Lword_Rword.
  • [0062]
    In the above example FIG. 3, the word extractor 140 applies word extraction rule 125-2 to example utterance 110-2 “I would like to speak with a customer representative.” For this utterance 110-2 and application of word extraction rule 125-2 FIG. 4) as shown, MVp Lword=“speak”, MVp Xword=“with”, Js Xword=“with”, Js Rword=“representative” to produce the candidate word grouping speak_representative based on format Lword_Rword.
  • [0063]
    As discussed below, a given utterance can include one candidate word grouping or multiple candidate word groupings.
  • [0064]
    FIG. 5 is an example diagram illustrating a listing of different example utterances that have been classified under a respective intent value (e.g., speak_representative) according to embodiments herein. That is, processing of each of the example utterances in FIG. 5 via the speech processing system 100 would produce the candidate word grouping 145-1 (e.g., a candidate intent value indicating a possible intended meaning of the utterance) for the respective utterance. For example, during processing, syntactic parser 115 parses a respective utterance to produce corresponding syntactic relationship information 120. The word extractor 140 then utilizes the respective word extraction rules 125 to identify that the respective utterance includes the respective candidate word grouping 145-1 speak_representative.
  • [0065]
    FIG. 6 is an example diagram illustrating a listing of different utterances that have been classified under a respective intent value (e.g., cancel_delivery) according to embodiments herein. That is, processing of the example utterances 110 in FIG. 6 via the speech processing system 100 would produce the candidate word grouping 145-2 (e.g., a candidate intent value indicating a possible intended meaning of the utterance) for the respective utterance. For example, during processing, syntactic parser 115 would parse the utterances to produce corresponding syntactic relationship information. The word extractor 140 then utilizes the respective word extraction rules 125 to identify that each of the utterances 110 as shown include the respective candidate word groupings 145-1.
  • [0066]
    FIG. 7 is an example diagram illustrating a listing of different utterances that have been classified under a respective intent value (e.g., pay_bill) according to embodiments herein. That is, processing of each of the example utterances 110 in FIG. 7 via the speech processing system 100 would produce the candidate word grouping 145-3 (e.g., a candidate intent value indicating a possible intended meaning of the utterance) for the respective utterance. During processing, syntactic parser 115 parses a respective utterance to produce corresponding syntactic relationship information. The word extractor 140 then utilizes the respective word extraction rules 125 to identify that the respective utterance includes the respective candidate word grouping 145-1 pay_bill.
  • [0067]
    FIG. 8 is an example diagram illustrating a listing of different utterances that have been classified under a respective intent value (e.g., when_delivery) according to embodiments herein. That is, processing of each of the example utterances 110 in FIG. 8 via the speech processing system 100 would produce the candidate word grouping 145-4 (e.g., a candidate intent value indicating a possible intended meaning of the utterance) for the utterance. During processing, syntactic parser 115 parses a respective utterance to produce corresponding syntactic relationship information. The word extractor 140 then utilizes the respective word extraction rules 125 to identify that the respective utterance includes the respective candidate word grouping 145-1 when_delivery.
  • [0068]
    FIG. 9 is an example diagram illustrating statistical information 160 indicating how often a respective intent value or candidate word grouping occurs in a pool of received utterances according to embodiments herein. For example, for a pool of received utterances, 457 of the utterances in the received pool included the intent value (or group of words) speak_representative, 337 of the utterances in the received pool included the intent value (or group of words) cancel_delivery, 312 of the utterances in the received pool included the intent value (or group of words) place_order, etc. Accordingly, embodiments herein can include keeping track of a frequency of occurrence for each of the different intent values for a pool of received utterances.
  • [0069]
    In one embodiment, the speech processing system 100 is unsupervised and requires no training data. The speech processing system 100 can collect and record the statistical information 160 over time as the speech processing system 100 receives and processes the additional utterances. Accordingly, embodiments herein include maintaining statistical information for a pool of previously received utterances. As previously discussed, the statistical information 160 indicate a frequency of receiving utterances of different intent types. The speech processing system 100 updates the statistical information 160 to reflect that the different utterances were assigned the intent values.
  • [0070]
    FIG. 10 is an example diagram illustrating intent values assigned to a respective subject matter label according to embodiments herein. Assume in this example that the speech processing system 100 receives the utterance “I would like to speak with a representative to change my order and cancel a delivery” in response to a query such as “How may I assist you?” In such an instance, the syntactic parser 115 processes the received utterance to produce respective syntactic relationship information 120 for the utterance.
  • [0071]
    In a manner as previously discussed, the word extractor 140 applies the word extraction rules 125 (such as those in FIG. 4) and syntactic relationship information 120 to identify candidate word groupings 145 such as speak_representative, change_order, and cancel_delivery as possible intended meanings of the utterance.
  • [0072]
    As previously discussed, the speech processing system 100 utilizes the identified syntactic relationships 120 of words to identify how the words in the received utterance are related. The speech processing system 100 then initiates application of word extraction rules or pattern rules to related words in the utterance to identify locations of words and produce candidate word groupings 145. As previously discussed, the word extraction rules 125 specifying which type of related words in the utterance to create the candidate sets of words.
  • [0073]
    In the example shown, the syntactic parser 115 produces respective syntactic relationship information 120 for the received utterance “I would like to speak with a representative to change my order and cancel a delivery”. By applying word extraction rules 125 to syntactic relationship information 120 and text in received utterance in a manner as previously discussed, the word extractor 140 produces a first set of words (e.g., one of candidate word groupings 145) to include a first word and a second word in the utterance such as “speak” and “representative”, the first word “speak” is syntactically related to the second word “representative” as indicated by the identified syntactic relationships 120 as previously discussed; the word extractor 140 produces a second set of words (e.g., another of candidate word groupings 145) to include a third word such as “change” and a fourth word “order” in the utterance, the third word “change” is syntactically related to the fourth word “order” as indicated by syntactic relationship information 120 for the received utterance; the word extractor 140 produces a third set of words (e.g., yet another of candidate word groupings 145) to include a fifth word such as “cancel” and a sixth word “delivery” in the utterance, the fifth word “cancel” is syntactically related to the sixth word “delivery” as indicated by syntactic relationship information 120 for the utterance.
  • [0074]
    As shown, the word extractor 140 maps the combination of words including “speak” and “representative” to possible intent value speak_representative; the word extractor 140 maps the combination of words including “change” and “order” to possible intent value change_order; the word extractor 140 maps the combination of words including “cancel” and “delivery” to possible intent value cancel_delivery.
  • [0075]
    Given these candidate word groupings 145 (namely, speak_representative, change_order, and cancel_delivery), the received utterance potentially can be classified in a number of different classes.
  • [0076]
    In one embodiment, the speech processing system 100 determines a dominant intent for assigning to the utterance based on statistical information 160 in FIG. 9. For example, the speech processing system 100 can determine how often the word groupings in the instance utterance under test appeared in other received utterances and select. The most often occurring intent value can be chosen for the instant utterance under test as the most likely intended meaning. For example, the intent value speak_representative appeared in 457 previously received utterances of a pool of received utterances, the intent value change_order appeared in 129 previously received utterances of the pool of received utterance, the intent value cancel_delivery appeared in 337 previously received utterances of a pool of received utterance.
  • [0077]
    In this example, based on the analysis, the analyzer 150 selects the candidate intent value speak_representative as being the most likely dominant intent for the utterance because it occurred most often in other previously received utterances. That is, in one example embodiment, the analyzer identifies a frequency of occurrence (e.g., number of utterances in a pool that include a specific grouping of words) that utterances in a pool of previously received utterances were of a same intent type as that of a first candidate intent value for the newly received utterance; the analyzer also identifies a frequency of occurrence that utterances in the pool of previously received utterances were of a same intent type as that of the second candidate intent value for the newly received utterance; and so on. The analyzer 150 then selects an intent value for the utterance under test “I would like to speak with a representative to change my order and cancel a delivery” based on the most often occurring intent value in previous utterances.
  • [0078]
    The analyzer 150 can be configured to tag the respective utterance depending on the one or more intent values (namely, speak_representative, change_order, and cancel_delivery) identified for the utterance. FIG. 11 illustrates how to map a respective intent value for an utterance to a respective label. Based on
  • [0079]
    FIG. 11 is an example diagram illustrating intent values and corresponding assigned subject matter labels according to embodiments herein. As shown, the label map can include multiple labels for potentially tagging the received utterance. Each of the labels can be assigned multiple word groupings that commonly appear in utterance that fall into the corresponding label. For example, the label “AGENT” is a candidate reserved for tagging any utterance including the word groupings speak-representative, speak_someone, talk_person, etc.; the label “SKIP_A_DELIVERY” can be reserved for tagging any utterance including the word groupings skip_delivery, skip_order, hold_delivery, etc.; the label “AGENT_BILLING” can be reserved for tagging any utterance including the word groupings have_bill, talk_bill, speak_billing, etc.; and so on.
  • [0080]
    Thus, the analyzer 150 can utilize the label map 1100 to identify how to label received utterances. Depending on the embodiment, an utterance can be assigned one or more labels indicating a class in to which the utterance falls. If desired, the example utterance which produces candidate word groupings speak_representative, change_order, and cancel_delivery can be assigned labels AGENT, CHANGE_ITEMS, AND CANCEL_DELIVERY. The utterance also can be labeled with only a single label corresponding to the dominant intent value (e.g., speak_representative) such as AGENT.
  • [0081]
    FIG. 12 is an example diagram illustrating possible association of one or more labels to an utterance according to embodiments herein.
  • [0082]
    For example, as previously discussed with respect to FIGS. 5-9, as shown in listing 1200, 457 utterances in a pool of previously received utterances included the intent value speak_representative. In 452 utterances of the 457 instances, the appropriate label for each of the utterances was the label “AGENT”. In 4 of the 457 instances, the appropriate label for each of the utterances was the label “AGENT_BILLING”. In 1 of the 457 instances, the appropriate label for respective utterance was the label “BILLING”.
  • [0083]
    Accordingly, the candidate word groupings 145 derived for an utterance can indicate that a respective utterance may fall under one or more of multiple different classes such as “AGENT” (a majority label for an utterance including speak_representative), “AGENT_BILLING” (a minority label for an utterance including speak_representative), and “BILLING” (a minority label for an utterance including speak_representative).
  • [0084]
    FIG. 13 is a diagram illustrating an example computer architecture for executing a speech processing system 100 according to embodiments herein.
  • [0085]
    Computer system 1300 can include one or more computerized devices such as a personal computer, workstation, portable computing device, console, network terminal, processing device, network device, etc., operating as a server, client, etc. The speech processing application 140-1 can be configured to include instructions to carry out any or all of the operations associated with syntactic parser 115, word extractor 120, analyzer 150, etc.
  • [0086]
    Note that the following discussion provides a basic embodiment indicating how to execute aspects of speech processing system 100 according to embodiments herein. However, it should be noted that the actual configuration for carrying out the operations as described herein can vary depending on a respective application.
  • [0087]
    As shown, computer system 1300 of the present example includes an interconnect 1311 that couples computer readable storage media 1312 such as a non-transitory type of media in which digital information can be stored and retrieved, a processor 1313, I/O interface 1314, and a communications interface 1317.
  • [0088]
    I/O interface 1314 enables receipt of utterances 110. I/O interface 1314 provides connectivity to repository 180 and, if present, other devices such as display screen, peripheral devices 316, keyboard, computer mouse, etc. Resources such as word extraction rules 125, statistical information 160, syntactic relationship information 120, candidate word groupings, etc. can be stored and retrieved from repository 180.
  • [0089]
    Computer readable storage medium 1312 can be any suitable device such as memory, optical storage, hard drive, floppy disk, etc. In one embodiment, the computer readable storage medium 1312 is a non-transitory storage media to store instructions and/or data.
  • [0090]
    Communications interface 1317 enables the computer system 1300 and processor 1313 to communicate over a network 190 to retrieve information from remote sources and communicate with other computers. I/O interface 1314 enables processor 1313 to retrieve or attempt retrieval of stored information from repository 180.
  • [0091]
    As shown, computer readable storage media 1312 can be encoded with speech processing application 140-1 (e.g., software, firmware, etc.) executed by processor 1313.
  • [0092]
    During operation of one embodiment, processor 1313 accesses computer readable storage media 1312 via the use of interconnect 1311 in order to launch, run, execute, interpret or otherwise perform the instructions of speech processing application 140-1 stored on computer readable storage medium 1312. As previously discussed, speech processing application 140-1 can include appropriate instructions, parsers, language models, analyzers, etc., to carry out any or all functionality associated with the speech processing system 100 as discussed herein.
  • [0093]
    Execution of the speech processing application 140-1 produces processing functionality such as speech processing process 140-2 in processor 1313. In other words, the speech processing process 140-2 associated with processor 1313 represents one or more aspects of executing speech processing application 140-1 within or upon the processor 1313 in the computer system 1300.
  • [0094]
    Those skilled in the art will understand that the computer system 1300 can include other processes and/or software and hardware components, such as an operating system that controls allocation and use of hardware resources to execute speech recognition application 140-1.
  • [0095]
    In accordance with different embodiments, note that computer system may be any of various types of devices, including, but not limited to, a personal computer system, desktop computer, laptop, notebook, netbook computer, mainframe computer system, handheld computer, workstation, network computer, application server, storage device, a consumer electronics device such as a camera, camcorder, set top box, mobile device, video game console, handheld video game device, a peripheral device such as a switch, modem, router, or in general any type of computing or electronic device.
  • [0096]
    Functionality supported by speech processing system 100 and speech processing application 140-1 will now be discussed via flowcharts in FIGS. 14-16. As discussed above, the speech recognition system 140 can be configured to execute the steps in the flowcharts as discussed below.
  • [0097]
    Note that there will be some overlap with respect to concepts discussed above. Also, note that the steps in the below flowcharts need not always be executed in the order shown. That is, the steps can be executed in any suitable order.
  • [0098]
    FIG. 14 is a flowchart 1400 illustrating a general technique of implementing a speech processing system 100 and related resources according to embodiments herein.
  • [0099]
    In step 1410, the speech processing system 100 parses an utterance 110-1 to identify syntactic relationships 120-1 amongst words in the utterance 110-1.
  • [0100]
    In step 1420, the speech processing system 100 groups or creates sets of words from the utterance 110-1 based on word extraction rules and the syntactic relationships of words in the utterance 110-1.
  • [0101]
    In step 1430, the speech processing system 100 maps each set of the sets of words (e.g., candidate word groupings 145) to a respective candidate intent value (e.g., possible intended meaning of the utterance).
  • [0102]
    In step 1440, the speech processing system 100 produces a list including candidate intent values for each of the sets of words (e.g., candidate word groupings 145).
  • [0103]
    In step 1450, the speech processing system 100 selects, from the list, a candidate intent value as being representative of an intent of the utterance.
  • [0104]
    FIGS. 15 and 16 combine to form a flowchart 1500 (e.g., flowchart 1500-1 and flowchart 1500-2) illustrating implementation of a speech processing system 100 according to embodiments herein.
  • [0105]
    In step 1510, the speech-processing system 100 parses text in a received utterance 110-1 to identify syntactic relationships amongst words in the utterance 110-1.
  • [0106]
    In step 1515, the speech-processing system 100 groups or creates sets of words from the received utterance based on word extraction rules 125 and/or the syntactic relationships (as specified by syntactic relationship information 120) of words in the utterance.
  • [0107]
    In sub-step 1520, the speech-processing system 100 utilizes the identified syntactic relationships amongst words to identify groupings (e.g., candidate word groupings 145) of related words in the utterance.
  • [0108]
    In sub-step 1525, the speech-processing system 100 applies a set of word extraction rules 125 and/or patterns to the identified syntactic relationships and syntactic relationship information 120 to identify locations of words in the utterance to create the sets of words.
  • [0109]
    In step 1530, the speech-processing system 100 maps each set of the sets of words to a respective candidate intent value.
  • [0110]
    In step 1535, the speech-processing system 100 produces a list including a candidate intent value for each of the sets of words. In one embodiment, the list includes a first candidate intent value, a second candidate intent value, and so on.
  • [0111]
    In step 1610, the speech-processing system 100 selects, from the list, a candidate intent value as being representative of an intent of the received utterance.
  • [0112]
    In sub-step 1615, the speech-processing system 100 identifies a frequency of occurrence that utterances in a pool of previously received utterances were of a same intent type as that of a first candidate intent value.
  • [0113]
    In step 1620, the speech-processing system 100 identifies a frequency of occurrence that utterances in the pool of previously received utterances were of a same intent type as that of the second candidate intent value.
  • [0114]
    In step 1625, the speech-processing system 100 selects the candidate intent value for the utterance depending on which of the first candidate intent value and the second candidate intent value occurred more soften in the pool for the previously received utterances. The selected candidate value indicates a dominant subject matter representative of the utterance.
  • [0115]
    In step 1630, the speech-processing system 100 identifies a tag representative of the selected candidate intent value for the utterance.
  • [0116]
    In step 1635, the speech-processing system 100 tags the utterance with the tag indicate a dominant subject matter intended by the utterance.
  • [0117]
    As discussed above, techniques herein are well suited for use in software and/or hardware applications implementing speech recognition and classification of utterances based on intended meanings. However, it should be noted that embodiments herein are not limited to use in such applications and that the techniques discussed herein are well suited for other applications
  • [0118]
    Based on the description set forth herein, numerous specific details have been set forth to provide a thorough understanding of claimed subject matter. However, it will be understood by those skilled in the art that claimed subject matter may be practiced without these specific details. In other instances, methods, apparatuses, systems, etc., that would be known by one of ordinary skill have not been described in detail so as not to obscure claimed subject matter. Some portions of the detailed description have been presented in terms of algorithms or symbolic representations of operations on data bits or binary digital signals stored within a computing system memory, such as a computer memory. These algorithmic descriptions or representations are examples of techniques used by those of ordinary skill in the data processing arts to convey the substance of their work to others skilled in the art. An algorithm as described herein, and generally, is considered to be a self-consistent sequence of operations or similar processing leading to a desired result. In this context, operations or processing involve physical manipulation of physical quantities. Typically, although not necessarily, such quantities may take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared or otherwise manipulated. It has proven convenient at times, principally for reasons of common usage, to refer to such signals as bits, data, values, elements, symbols, characters, terms, numbers, numerals or the like. It should be understood, however, that all of these and similar terms are to be associated with appropriate physical quantities and are merely convenient labels. Unless specifically stated otherwise, as apparent from the following discussion, it is appreciated that throughout this specification discussions utilizing terms such as “processing,” “computing,” “calculating,” “determining” or the like refer to actions or processes of a computing platform, such as a computer or a similar electronic computing device, that manipulates or transforms data represented as physical electronic or magnetic quantities within memories, registers, or other information storage devices, transmission devices, or display devices of the computing platform.
  • [0119]
    While this invention has been particularly shown and described with references to preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present application as defined b the appended claims. Such variations are intended to be covered by the scope of this present application. As such, the foregoing description of embodiments of the present application is not intended to be limiting. Rather, any limitations to the invention are presented in the following claims.
Patentzitate
Zitiertes PatentEingetragen Veröffentlichungsdatum Antragsteller Titel
US5077804 *11. Dez. 199031. Dez. 1991Richard Dnaiel DTelecommunications device and related method
US5278980 *16. Aug. 199111. Jan. 1994Xerox CorporationIterative technique for phrase query formation and an information retrieval system employing same
US5418717 *12. Dez. 199123. Mai 1995Su; Keh-YihMultiple score language processing system
US5457768 *12. Aug. 199210. Okt. 1995Kabushiki Kaisha ToshibaSpeech recognition apparatus using syntactic and semantic analysis
US5597312 *4. Mai 199428. Jan. 1997U S West Technologies, Inc.Intelligent tutoring method and system
US5625748 *18. Apr. 199429. Apr. 1997Bbn CorporationTopic discriminator using posterior probability or confidence scores
US5671329 *9. März 199423. Sept. 1997Nec CorporationSpeech dialogue system in which a recognition and understanding process, application process, and voice input response are performed simultaneously with voice input
US5867817 *19. Aug. 19962. Febr. 1999Virtual Vision, Inc.Speech recognition manager
US5937384 *1. Mai 199610. Aug. 1999Microsoft CorporationMethod and system for speech recognition using continuous density hidden Markov models
US5966686 *28. Juni 199612. Okt. 1999Microsoft CorporationMethod and system for computing semantic logical forms from syntax trees
US6178398 *18. Nov. 199723. Jan. 2001Motorola, Inc.Method, device and system for noise-tolerant language understanding
US6185531 *9. Jan. 19986. Febr. 2001Gte Internetworking IncorporatedTopic indexing method
US6223150 *29. Jan. 199924. Apr. 2001Sony CorporationMethod and apparatus for parsing in a spoken language translation system
US6236968 *14. Mai 199822. Mai 2001International Business Machines CorporationSleep prevention dialog based car system
US6243669 *29. Jan. 19995. Juni 2001Sony CorporationMethod and apparatus for providing syntactic analysis and data structure for translation knowledge in example-based language translation
US6442524 *29. Jan. 199927. Aug. 2002Sony CorporationAnalyzing inflectional morphology in a spoken language translation system
US6519562 *25. Febr. 199911. Febr. 2003Speechworks International, Inc.Dynamic semantic control of a speech recognition system
US6697793 *2. März 200124. Febr. 2004The United States Of America As Represented By The Administrator Of The National Aeronautics And Space AdministrationSystem, method and apparatus for generating phrases from a database
US6766320 *24. Aug. 200020. Juli 2004Microsoft CorporationSearch engine with natural language-based robust parsing for user query and relevance feedback learning
US6816831 *27. Okt. 20009. Nov. 2004Sony CorporationLanguage learning apparatus and method therefor
US6937975 *22. Sept. 199930. Aug. 2005Canon Kabushiki KaishaApparatus and method for processing natural language
US6961954 *2. März 19981. Nov. 2005The Mitre CorporationAutomated segmentation, information extraction, summarization, and presentation of broadcast news
US7006973 *31. Jan. 200028. Febr. 2006Intel CorporationProviding information in response to spoken requests
US7085708 *18. Juni 20011. Aug. 2006Ravenflow, Inc.Computer system with natural language to machine language translator
US7089226 *28. Juni 20018. Aug. 2006Microsoft CorporationSystem, representation, and method providing multilevel information retrieval with clarification dialog
US7177817 *12. Dez. 200213. Febr. 2007Tuvox IncorporatedAutomatic generation of voice content for a voice response system
US7313523 *14. Mai 200325. Dez. 2007Apple Inc.Method and apparatus for assigning word prominence to new or previous information in speech synthesis
US7739103 *6. Apr. 200515. Juni 2010Educational Testing ServiceLexical association metric for knowledge-free extraction of phrasal terms
US7912701 *4. Mai 200722. März 2011IgniteIP Capital IA Special Management LLCMethod and apparatus for semiotic correlation
US7912702 *31. Okt. 200722. März 2011Phoenix Solutions, Inc.Statistical language model trained with semantic variants
US7983896 *3. März 200519. Juli 2011SDL Language TechnologyIn-context exact (ICE) matching
US7996219 *21. März 20059. Aug. 2011At&T Intellectual Property Ii, L.P.Apparatus and method for model adaptation for spoken language understanding
US8255383 *13. Juli 200728. Aug. 2012Chacha Search, IncMethod and system for qualifying keywords in query strings
US8311807 *9. Nov. 200513. Nov. 2012Samsung Electronics Co., Ltd.Periodically extracting and evaluating frequency of occurrence data of unregistered terms in a document for updating a dictionary
US8321220 *30. Nov. 200527. Nov. 2012At&T Intellectual Property Ii, L.P.System and method of semi-supervised learning for spoken language understanding using semantic role labeling
US8370352 *15. Okt. 20085. Febr. 2013Siemens Medical Solutions Usa, Inc.Contextual searching of electronic records and visual rule construction
US8521526 *28. Juli 201027. Aug. 2013Google Inc.Disambiguation of a spoken query term
US8527262 *22. Juni 20073. Sept. 2013International Business Machines CorporationSystems and methods for automatic semantic role labeling of high morphological text for natural language processing applications
US8560321 *5. Jan. 201115. Okt. 2013Interactions CorportionAutomated speech recognition system for natural language understanding
US9002725 *29. Aug. 20057. Apr. 2015Google Inc.System and method for targeting information based on message content
US9093073 *12. Febr. 200728. Juli 2015West CorporationAutomatic speech recognition tagging
US9177045 *2. Juni 20103. Nov. 2015Microsoft Technology Licensing, LlcTopical search engines and query context models
US9317593 *1. Okt. 200819. Apr. 2016Fujitsu LimitedModeling topics using statistical distributions
US9330168 *13. Febr. 20143. Mai 2016Go Daddy Operating Company, LLCSystem and method for identifying website verticals
US20010047262 *5. Febr. 200129. Nov. 2001Alexander KurganovRobust voice browser system and voice activated device controller
US20020002454 *26. Juli 20013. Jan. 2002Srinivas BangaloreAutomatic clustering of tokens from a corpus for grammar acquisition
US20020029304 *17. Juli 20017. März 2002Microsoft CorporationMethod and system for defining semantic categories and actions
US20020032561 *7. Sept. 200114. März 2002Nec CorporationAutomatic interpreting system, automatic interpreting method, and program for automatic interpreting
US20020059066 *8. Apr. 199816. Mai 2002O'hagan Timothy P.Speech recognition system and method for employing the same
US20020128821 *11. März 200212. Sept. 2002Farzad EhsaniPhrase-based dialogue modeling with particular application to creating recognition grammars for voice-controlled user interfaces
US20020169596 *4. Mai 200114. Nov. 2002Brill Eric D.Method and apparatus for unsupervised training of natural language processing units
US20030033333 *22. März 200213. Febr. 2003Fujitsu LimitedHot topic extraction apparatus and method, storage medium therefor
US20030105638 *27. Nov. 20015. Juni 2003Taira Rick K.Method and system for creating computer-understandable structured medical data from natural language reports
US20030108334 *6. Dez. 200112. Juni 2003Koninklijke Philips Elecronics N.V.Adaptive environment system and method of providing an adaptive environment
US20030163321 *18. Juni 200128. Aug. 2003Mault James RSpeech recognition capability for a personal digital assistant
US20030233224 *19. Febr. 200318. Dez. 2003Insightful CorporationMethod and system for enhanced data searching
US20040059577 *26. Juni 200325. März 2004International Business Machines CorporationMethod and apparatus for preparing a document to be read by a text-to-speech reader
US20040078190 *28. Sept. 200122. Apr. 2004Fass Daniel CMethod and system for describing and identifying concepts in natural language text for information retrieval and processing
US20040148170 *30. Mai 200329. Juli 2004Alejandro AceroStatistical classifiers for spoken language understanding and command/control scenarios
US20040158558 *22. Juli 200312. Aug. 2004Atsuko KoizumiInformation processor and program for implementing information processor
US20050049867 *11. Aug. 20043. März 2005Paul DeaneCooccurrence and constructions
US20050055209 *5. Sept. 200310. März 2005Epstein Mark E.Semantic language modeling and confidence measurement
US20050080629 *16. Jan. 200314. Apr. 2005David AttwaterMulti-mode interactive dialogue apparatus and method
US20050105712 *10. Sept. 200419. Mai 2005Williams David R.Machine learning
US20050144000 *22. Dez. 200430. Juni 2005Kabushiki Kaisha ToshibaContents providing apparatus and method
US20060056602 *13. Sept. 200416. März 2006Sbc Knowledge Ventures, L.P.System and method for analysis and adjustment of speech-enabled systems
US20060074671 *5. Okt. 20046. Apr. 2006Gary FarmanerSystem and methods for improving accuracy of speech recognition
US20060080098 *29. Sept. 200513. Apr. 2006Nick CampbellApparatus and method for speech processing using paralinguistic information in vector form
US20060106596 *29. Dez. 200518. Mai 2006Microsoft CorporationRanking Parser for a Natural Language Processing System
US20060129397 *10. Dez. 200415. Juni 2006Microsoft CorporationSystem and method for identifying semantic intent from acoustic information
US20060173683 *3. Febr. 20063. Aug. 2006Voice Signal Technologies, Inc.Methods and apparatus for automatically extending the voice vocabulary of mobile communications devices
US20060173686 *1. Febr. 20063. Aug. 2006Samsung Electronics Co., Ltd.Apparatus, method, and medium for generating grammar network for use in speech recognition and dialogue speech recognition
US20060197764 *2. März 20057. Sept. 2006Yang George LDocument animation system
US20060235843 *31. Jan. 200619. Okt. 2006Textdigger, Inc.Method and system for semantic search and retrieval of electronic documents
US20060259294 *16. Dez. 200316. Nov. 2006John TashereauVoice recognition system and method
US20060259299 *26. Dez. 200316. Nov. 2006Yumiko KatoBroadcast reception method, broadcast reception systm, recording medium and program (as amended)
US20070005369 *30. Juni 20054. Jan. 2007Microsoft CorporationDialog analysis
US20070118374 *25. Okt. 200624. Mai 2007Wise Gerald BMethod for generating closed captions
US20070136048 *13. Dez. 200514. Juni 2007David Richardson-BunburySystem for classifying words
US20070156747 *12. Dez. 20065. Juli 2007Tegic Communications LlcMobile Device Retrieval and Navigation
US20070174057 *28. März 200726. Juli 2007Genly Christopher HProviding programming information in response to spoken requests
US20070225980 *1. März 200727. Sept. 2007Kabushiki Kaisha ToshibaApparatus, method and computer program product for recognizing speech
US20080046244 *1. Nov. 200521. Febr. 2008Yoshio OhnoSpeech Recognition Device
US20080071536 *11. Sept. 200720. März 2008Honda Motor Co., Ltd.Voice recognition device, voice recognition method, and voice recognition program
US20080140389 *4. Dez. 200712. Juni 2008Honda Motor Co., Ltd.Language understanding apparatus, language understanding method, and computer program
US20080141125 *22. Juni 200712. Juni 2008Firooz GhassabianCombined data entry systems
US20080154870 *12. Febr. 200726. Juni 2008Voice Signal Technologies, Inc.Collection and use of side information in voice-mediated mobile search
US20080201136 *18. Sept. 200721. Aug. 2008Kabushiki Kaisha ToshibaApparatus and Method for Speech Recognition
US20080221903 *22. Mai 200811. Sept. 2008International Business Machines CorporationHierarchical Methods and Apparatus for Extracting User Intent from Spoken Utterances
US20080243820 *27. März 20072. Okt. 2008Walter ChangSemantic analysis documents to rank terms
US20080319748 *31. Jan. 200725. Dez. 2008Mikio NakanoConversation System and Conversation Software
US20090006343 *28. Juni 20071. Jan. 2009Microsoft CorporationMachine assisted query formulation
US20090006345 *28. Juni 20071. Jan. 2009Microsoft CorporationVoice-based search processing
US20090063426 *29. Aug. 20085. März 2009Powerset, Inc.Identification of semantic relationships within reported speech
US20090077047 *14. Aug. 200619. März 2009Inquira, Inc.Method and apparatus for identifying and classifying query intent
US20090094233 *1. Okt. 20089. Apr. 2009Fujitsu LimitedModeling Topics Using Statistical Distributions
US20090150156 *11. Dez. 200711. Juni 2009Kennewick Michael RSystem and method for providing a natural language voice user interface in an integrated voice navigation services environment
US20090209345 *23. Jan. 200920. Aug. 2009Aruze Gaming America, Inc.Multiplayer participation type gaming system limiting dialogue voices outputted from gaming machine
US20090259650 *11. Apr. 200815. Okt. 2009Ebay Inc.System and method for identification of near duplicate user-generated content
US20090276419 *1. Mai 20095. Nov. 2009Chacha Search Inc.Method and system for improvement of request processing
US20090285474 *15. Mai 200819. Nov. 2009Berteau Stefan ASystem and Method for Bayesian Text Classification
US20090313227 *11. Juni 200917. Dez. 2009Veoh Networks, Inc.Searching Using Patterns of Usage
US20090327260 *25. Juni 200831. Dez. 2009Microsoft CorporationConstructing a classifier for classifying queries
US20100023506 *31. Dez. 200828. Jan. 2010Yahoo! Inc.Augmenting online content with additional content relevant to user interests
US20100057687 *4. Sept. 20084. März 2010Microsoft CorporationPredicting future queries from log data
US20100094854 *9. Febr. 200915. Apr. 2010Omid Rouhani-KallehSystem for automatically categorizing queries
US20100114908 *4. Nov. 20086. Mai 2010Microsoft CorporationRelevant navigation with deep links into query
US20100121840 *12. Nov. 200813. Mai 2010Yahoo! Inc.Query difficulty estimation
US20100131835 *13. Nov. 200927. Mai 2010Srihari KumarSystem and methods for inferring intent of website visitors and generating and packaging visitor information for distribution as sales leads or market intelligence
US20100138402 *2. Dez. 20093. Juni 2010Chacha Search, Inc.Method and system for improving utilization of human searchers
US20100145710 *8. Dez. 200810. Juni 2010Nuance Communications, Inc.Data-Driven Voice User Interface
US20100153106 *15. Dez. 200817. Juni 2010Verizon Data Services LlcConversation mapping
US20100153317 *24. Nov. 200917. Juni 2010Samsung Electronics Co., LtdIntelligent robot and control method thereof
US20100174716 *15. März 20108. Juli 2010Google Inc.Methods and systems for improving text segmentation
US20100228762 *5. März 20109. Sept. 2010Mauge KarinSystem and method to provide query linguistic service
US20100268536 *17. Apr. 200921. Okt. 2010David SuendermannSystem and method for improving performance of semantic classifiers in spoken dialog systems
US20100293174 *12. Mai 200918. Nov. 2010Microsoft CorporationQuery classification
US20110004462 *1. Juli 20096. Jan. 2011Comcast Interactive Media, LlcGenerating Topic-Specific Language Models
US20110010367 *11. Juni 201013. Jan. 2011Chacha Search, Inc.Method and system of providing a search tool
US20110029311 *17. Juni 20103. Febr. 2011Sony CorporationVoice processing device and method, and program
US20110029533 *28. Juli 20093. Febr. 2011Prasantha JayakodyMethod and system for tag suggestion in a tag-associated data-object storage system
US20110099003 *13. Sept. 201028. Apr. 2011Masaaki IsozuInformation processing apparatus, information processing method, and program
US20110184730 *22. Jan. 201028. Juli 2011Google Inc.Multi-dimensional disambiguation of voice commands
US20110191423 *29. Jan. 20104. Aug. 2011Mcafee, Inc.Reputation management for network content classification
US20110213777 *3. Sept. 20101. Sept. 2011Alibaba Group Holding LimitedMethod and Apparatus of Text Classification
US20110225019 *14. Okt. 200915. Sept. 2011David TaylorSearch, analysis and categorization
US20110238408 *26. März 201029. Sept. 2011Jean-Marie Henri Daniel LarchevequeSemantic Clustering
US20110238410 *26. März 201029. Sept. 2011Jean-Marie Henri Daniel LarchevequeSemantic Clustering and User Interfaces
US20110258204 *30. Juni 201120. Okt. 2011Wordnetworks, Inc.System for using keyword phrases on a page to provide contextually relevant content to users
US20110282913 *15. Dez. 200917. Nov. 2011Oki Electric Industry Co., Ltd.Dialogue control system, method and computer readable storage medium, and multidimensional ontology processing system, method and computer readable storage medium
US20110288868 *19. Mai 201024. Nov. 2011Lloyd Matthew IDisambiguation of contact information using historical data
US20110301943 *9. Aug. 20118. Dez. 2011Redstart Systems, Inc.System and method of dictation for a speech recognition command system
US20110301955 *7. Juni 20108. Dez. 2011Google Inc.Predicting and Learning Carrier Phrases for Speech Input
US20110314012 *16. Juni 201022. Dez. 2011Microsoft CorporationDetermining query intent
US20110314390 *18. Juni 201022. Dez. 2011Microsoft CorporationTechniques to dynamically modify themes based on messaging
US20120016873 *16. Juli 201019. Jan. 2012Michael MathiesonMethod and system for ranking search results based on categories
US20120053927 *1. Sept. 20101. März 2012Microsoft CorporationIdentifying topically-related phrases in a browsing sequence
US20120078919 *29. Aug. 201129. März 2012Fujitsu LimitedComparison of character strings
US20120084291 *30. Sept. 20105. Apr. 2012Microsoft CorporationApplying search queries to content sets
US20120096033 *14. Okt. 201019. Apr. 2012Microsoft CorporationDisambiguation of Entities
US20120158693 *17. Dez. 201021. Juni 2012Yahoo! Inc.Method and system for generating web pages for topics unassociated with a dominant url
US20120158703 *16. Dez. 201021. Juni 2012Microsoft CorporationSearch lexicon expansion
US20120191453 *30. März 201226. Juli 2012Cyberpulse L.L.C.System and methods for matching an utterance to a template hierarchy
US20120290293 *16. Sept. 201115. Nov. 2012Microsoft CorporationExploiting Query Click Logs for Domain Detection in Spoken Language Understanding
US20120310628 *30. Mai 20126. Dez. 2012Samsung Electronics Co., Ltd.Method and system for providing access to information of potential interest to a user
US20120317107 *23. Jan. 201213. Dez. 2012Google Inc.Methods and Systems for Classifying Search Results to Determine Page Elements
US20130006995 *10. Dez. 20103. Jan. 2013Chesterdeal LimitedAccessing stored electronic resources
US20130138641 *30. Dez. 200930. Mai 2013Google Inc.Construction of text classifiers
US20130212475 *31. Okt. 201115. Aug. 2013Koninklijke Philips Electronics N.V.Suggesting relevant terms during text entry
US20140143054 *24. Jan. 201422. Mai 2014Yahoo! Inc.System for identifying and selecting advertising categories
US20140149399 *22. Juli 201029. Mai 2014Google Inc.Determining user intent from query patterns
Referenziert von
Zitiert von PatentEingetragen Veröffentlichungsdatum Antragsteller Titel
US8983840 *19. Juni 201217. März 2015International Business Machines CorporationIntent discovery in audio or text-based conversation
US96201473. Febr. 201511. Apr. 2017International Business Machines CorporationIntent discovery in audio or text-based conversation
Klassifizierungen
US-Klassifikation704/254
Internationale KlassifikationG10L15/00
UnternehmensklassifikationG06F17/2785, G06F17/271, G06F17/30743, G10L15/00, G10L15/1822
Juristische Ereignisse
DatumCodeEreignisBeschreibung
20. Febr. 2014ASAssignment
Owner name: NUANCE COMMUNICATIONS, INC., MASSACHUSETTS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:JOSHI, SACHINDRA;GODBOLE, SHANTANU;REEL/FRAME:032258/0913
Effective date: 20110211