WO2013143429A1 - 搜索引擎的推荐搜索方法、装置及计算机可读存储介质 - Google Patents

搜索引擎的推荐搜索方法、装置及计算机可读存储介质 Download PDF

Info

Publication number
WO2013143429A1
WO2013143429A1 PCT/CN2013/073129 CN2013073129W WO2013143429A1 WO 2013143429 A1 WO2013143429 A1 WO 2013143429A1 CN 2013073129 W CN2013073129 W CN 2013073129W WO 2013143429 A1 WO2013143429 A1 WO 2013143429A1
Authority
WO
WIPO (PCT)
Prior art keywords
search
query word
weight
vertical search
vertical
Prior art date
Application number
PCT/CN2013/073129
Other languages
English (en)
French (fr)
Inventor
何军
匡健锋
泮华杰
Original Assignee
腾讯科技(深圳)有限公司
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 腾讯科技(深圳)有限公司 filed Critical 腾讯科技(深圳)有限公司
Publication of WO2013143429A1 publication Critical patent/WO2013143429A1/zh
Priority to US14/497,993 priority Critical patent/US9934312B2/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/953Querying, e.g. by the use of web search engines
    • G06F16/9532Query formulation
    • 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
    • G06F16/33Querying
    • G06F16/332Query formulation
    • G06F16/3322Query formulation using system suggestions
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/903Querying
    • G06F16/9032Query formulation
    • G06F16/90324Query formulation using system suggestions
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/951Indexing; Web crawling techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/953Querying, e.g. by the use of web search engines
    • G06F16/9535Search customisation based on user profiles and personalisation

Definitions

  • the present invention relates to Internet information search technology, and in particular, to a search search method, a device, and a computer readable storage medium. Background technique
  • Internet search engine refers to collecting information from the Internet according to certain strategies and using specific programs. After organizing and processing the information, it provides search services for users. The search engine retrieves relevant information based on the keywords input by the user, and displays the retrieved related information as a search result to the user.
  • search engine technology An important goal of the development of search engine technology is to minimize the number of human-machine operations of users, and to try to show users the search results that match the user's search intention. To achieve this goal, search engine technology is constantly being improved and developed.
  • search engine recommendation query word display technology there is a search engine recommendation query word display technology.
  • the main processing process of this technology is: In the process of inputting query words in the search search box of the search engine, the search engine will query words according to the user input.
  • the query word abc finds the candidate query words that hit the abc through the text index, and then filters the candidate words by statistical data such as the number of queries and/or the click rate to obtain the query words that are finally recommended to the user (in the industry, such recommendation)
  • the query word is also referred to as the recommended query word), and the recommended query words finally recommended to the user are displayed in real time, thereby helping the user to filter out the query words that may be of interest, saving the time for the user to input the final query word, and improving the search efficiency.
  • FIG. 1 is an interface diagram of a search engine automatically calculating and displaying a recommended query word when a user inputs a query word in the prior art.
  • the user inputs a query word "Wang Fei" 101 in the search box, and the search engine
  • the recommendation calculation process is immediately performed to obtain a recommended query word, and a list 102 including the recommended query words is displayed.
  • the user may click on the recommended query word to complete a web search.
  • the search result of the webpage cannot satisfy the vertical search requirement of the user.
  • the so-called vertical search is a professional search technology for a certain industry. It is a subdivision and extension of the search engine. It is a special type of information in the web library. Once integrated, the directed subfield extracts the required data for processing and returns it to the user in some form.
  • a new search engine technology approach such as a large amount of information, inaccurate queries, and insufficient depth, compared to a general search engine, with valuable information provided by a specific domain, a specific group of people, or a specific need.
  • Related Services Its characteristics are "specialized, precise, deep", and has an industry color. Compared with the massive information disorder of the general search engine, the vertical search engine is more focused, specific and in-depth.
  • the web search engine of the current search engine is a general-purpose, comprehensive search technology, and the search result based on the query word contains the results of various sub-types of information, for example, the search result page includes videos and pictures. , news, music, and other types of search results.
  • Vertical search requires different types of information to be distinguished.
  • a vertical search engine only searches for one type of content.
  • the video vertical search engine is specifically used to search for the results of the video class
  • the news vertical search engine is specifically used to search the search results of the news category.
  • Most of the current search engines have different vertical search engines (also referred to as vertical search channels in the industry), which respectively search for different types of vertical search types.
  • FIG. 2 is a schematic diagram of an interface of a search engine homepage, including a webpage search engine 201 (ie, a general search engine), and also includes a photo search engine 202, a video search engine 203, a music search engine 204, and a question and answer search engine. 205 (ie, "question" in the figure), news search engine 206, and the like, such a vertical search engine.
  • a webpage search engine 201 ie, a general search engine
  • 205 ie, "question” in the figure
  • news search engine 206 ie, "question" in the figure
  • the like such a vertical search engine.
  • the prior art described in Fig. 1 saves the user's input time, but does not satisfy the user's vertical search requirements. For example, if the user clicks on the recommended query word "Wang Fei Legend” 103 or "Wang Fei Xi'an Concert" 104 in Figure 1, the corresponding web search result will be directly displayed. However, the actual search intent of different recommended query terms mostly corresponds to the corresponding vertical search type. For example, the actual search intent of "Wang Fei Legend" 103 mostly corresponds to the music content, and the actual search intention of "Wang Fei Xi'an Concert" 104 is mostly Corresponding to the video content. In the prior art shown in FIG. 1 , a more intuitive and highly relevant vertical search category cannot be separately listed. If the corresponding vertical search content is queried, the user also needs to perform a secondary search related vertical search channel link. In order to find the corresponding vertical search content, such as music content or video content.
  • the prior art has low search efficiency in vertical search, and it is inconvenient for the user to find a vertical search result with high relevance to the query word from the search result, and the user cannot know the vertical search with the highest correlation with each recommended query word.
  • Type At the same time, in order to select the final vertical search result, the user often needs to click the relevant vertical search channel link twice, resulting in the user and the person on the Internet machine side. The number of machine interactions increases, and each human-computer interaction operation issues operation request information, triggers the calculation process, and generates response result information, which occupies a large amount of resources on the machine side, including client resources, server resources, network bandwidth resources, and the like.
  • the main purpose of the embodiments of the present invention is to provide a search search method, a device, and a computer readable storage medium to improve the search efficiency of the vertical search.
  • a search engine recommendation search method including:
  • the vertical search engine corresponding to the vertical search type is used to search for the recommended query word to display the search result.
  • a search engine recommendation search device comprising:
  • a logging module configured to record a search log and a click log of a query word in the search process
  • a preference analysis module configured to analyze a vertical search tendency weight of the query word according to the recorded log
  • a query word recommendation module configured to detect a query word in the search box, and determine a recommended query word related to the query word after detecting the query word;
  • a vertical search propensity recommendation module configured to query a vertical search propensity weight corresponding to each recommended query word, and determine a propensity vertical search type of the recommended query word according to the preference weight;
  • a recommendation display module configured to display the recommended recommendations a query term and a link to its preferred vertical search type;
  • a response module configured to search for the recommended query word by using a vertical search engine corresponding to the vertical search type after detecting that the link of the preferred vertical search type corresponding to the recommended query word is clicked Match the content and display the search results.
  • a computer readable storage medium having stored thereon a set of instructions, wherein when the set of instructions is executed, the computer can execute a recommended search method of any one of the above search engines
  • FIG. 1 is an interface diagram of a search engine automatically calculating and displaying a recommended query word when a user inputs a query word in the prior art;
  • FIG. 2 is a schematic diagram of an interface of a conventional search engine homepage
  • FIG. 3 is a flowchart of processing of a recommendation search method of a search engine according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a composition search apparatus of a search engine according to an embodiment of the present invention.
  • FIG. 5 is an interface diagram of the recommendation search device automatically analyzing and displaying the recommended query words and their vertical search propensity types when the user inputs the query words into the search box according to an embodiment of the present invention
  • FIG. 6 is an embodiment of the present invention
  • a schematic diagram of search results after the user clicks on the preferred vertical search type link corresponding to the recommended query word
  • FIG. 7 is a schematic diagram of an interface of video class search results displayed in webpage search results. detailed description
  • FIG. 3 is a process flow diagram of a recommendation search method of a search engine according to an embodiment of the present invention.
  • the method mainly includes:
  • Step 301 Record a search log and a click log of the query word in the search process.
  • Step 302 Analyze a vertical search tendency weight of the query word according to the recorded log.
  • Step 303 After detecting the query word in the search box, determine a push related to the query word. Referring to the query word, querying the vertical search propensity weight corresponding to each recommended query word and determining the propensity vertical search type of the recommended query word according to the preference weight, and displaying the recommended search words and their tendency vertical search types link.
  • Step 304 After detecting that the link of the propensity vertical search type corresponding to the recommended query word is clicked, search for the matching content of the recommended query word by using a vertical search engine corresponding to the vertical search type, and display the search result.
  • FIG. 5 is a schematic diagram of the recommended search device automatically analyzing and displaying the recommended query words and their vertical search propensity types when the user inputs a query word into the search box according to an embodiment of the present invention. As shown in FIG.
  • the recommendation search device when the user inputs the query word 501 in the input box of the search engine, automatically determines a set of query words 502 associated with the query word input by the user. And displaying the set of query words 502 below the search box, and displaying a link 503 of the corresponding propensity vertical search type for the query words with the explicit preference is displayed next to the corresponding query words, for example, as shown in FIG. Displayed after the symbol @ after the corresponding query word.
  • the user can not only click on one of the recommended query words to perform webpage search, but also directly click on the propensity vertical search type link behind the symbol@ to directly call the corresponding type of vertical search engine to search for the corresponding query word, for example,
  • the user can click on the vertical search type link "@Video" corresponding to the query word "Wang Fei Xi'an Concert” in Figure 5, and then call the vertical search engine of the video class to search for the query word "Wang Fei Xi'an Concert” and will return as The video search results shown in Figure 6.
  • the process of searching for a vertical resource of interest in a large number of webpage search results can be saved, and the search result more in line with the user's demand can be obtained, and the search efficiency is improved.
  • the example of the interface shown in FIG. 5 and FIG. 6 is described by taking a computer search engine as an example.
  • the embodiment of the present invention can also be applied to a wireless search engine.
  • the steps described in the embodiment of the present invention are described in more detail.
  • the search log and the click log of the query word in the record search process in step 301 are performed in the user search process.
  • the search process refers to a search process before the recommendation query word and its tendency vertical search type described in the embodiment of the present invention, and may also be referred to as a relative history search process.
  • the search process mainly includes: the user submits a search request when using the search engine, including the query word, and the front-end server obtains the search request and sends a search request to the back-end system of the search engine, and the search engine returns a search to the user after searching. List of results. If the user finds a search result of interest, he or she clicks on one or several of the search results links to view it.
  • the logging module of the embodiment of the present invention needs to record two kinds of logs: one is a search log, and the other is a click log.
  • the search log is mainly for each type of vertical search engine, and the specific method for recording the search log of the query word is: for each type of vertical search engine, recording the query word input every search in the vertical search engine of the type Content, the number of links clicked in the search results, and the type of the vertical search engine;
  • the click log is mainly for a webpage search engine, and the specific method for recording the click log of the query word is: recording the content of the query word input in each search in the webpage search engine, the clicked link in the search result, the clicked link Corresponding vertical search type.
  • the two log data are input to the tilt analysis module for analysis and calculation, and the method of step 302 is used to obtain a biased query word recommendation set.
  • Step 302 analyzing the vertical search propensity weight of the query word according to the recorded log, specifically determining the preference weight of each query word for each vertical search type, wherein the preference for a certain search word for a certain vertical search type
  • the specific method of weight includes the following steps 321 to 323. The following is an example of the preference weight of the video type vertical search by the query word "Wang Fei Xi'an Concert":
  • Step 321 Analyze, according to the search log, the first preference weight pwl of the query word "Wang Fei Xi'an Concert" for the vertical search of the video class, and the specific determining method includes: determining a query weight X/Y, where X is specified The number of queries of the query word in the vertical search engine of the video class, Y is the total number of queries of the vertical search engine of the video class in the specified time period; determining the click weight N/M, where N is the The query term "Wang Fei Xi'an within the specified time period” The concert "the number of links that are clicked in the search results of the video-based vertical search engine, and M is the total search result of the video-based vertical search engine page within the specified time period (ie, all the query words of the video-based vertical search engine) The number of links clicked in the search result; weighting the query weight and the click weight to obtain the first tendency weight pw l of the query word "Wang Fei Xi'an Concert" for the video class
  • the specific manner of weighting the query weight and the click weight to obtain the first preference weight of the query word for the vertical search type may be obtained by using a weighted summation method, or may be obtained by using a product form.
  • the calculating the first propensity weight pwl by using the weighted summation is specifically: using the formula ( ⁇ ⁇ / ⁇ + ⁇ ⁇ / ⁇ ) to obtain the first propensity weight pwl, where ⁇ is the query weight factor, and ⁇ is Click on the weight factor, these two factors can be preset.
  • the calculating the first propensity weight pwl by using the product form is specifically: using a formula (1+X/Y) (1 + N/M) and normalizing to a floating point number of the (0, 1) interval, the floating The number of points is the first tendency weight pw l.
  • Step 322 Analyze, according to the click log, the second tendency weight pw2 of the query word "Wang Fei Xi'an Concert" for video-based vertical search, the second preference weight pw2 is y/x, where X is the query word The number of queries in the web search engine within a specified time period, for example, within a specified time period t, the user searches for X times of "Wang Fei Xi'an concert" in the web search engine page; y is clicked in the web search result of the query word The number of links corresponding to the vertical search of the video class, for example, the user clicks on the search result of the y video class in the search result for the query word "Wang Fei Xi'an Concert".
  • Step 323 Multiply the first preference weight pwl and the second preference weight pw2 to obtain a preference weight of the query word for the vertical search type.
  • a threshold of a preference weight may be set for filtering the preference weight. If the preference weight of a certain vertical search type of a query word is lower than the set threshold, If the preference is used, the preference weight of the vertical search type corresponding to the query word is 0. If the orientation weight of each type of vertical search type corresponding to the query word is 0, the query word in Table 1 corresponds to The propensity weight can be set to zero. If the preference weight of a certain type of vertical search type of a query word is greater than the threshold value, it indicates that the query word has a certain tendency to the vertical search type of the query word, and the higher the preference weight score, the higher the tendency .
  • the above query words are indexed and loaded into the memory by the recommended search device of the embodiment of the present invention.
  • the query word recommendation module of the embodiment of the present invention When it is detected that the user inputs the query word "Wang Fei" in the search box, the query word recommendation module of the embodiment of the present invention first finds the candidate word hitting Faye Wong through the text index, and then performs the candidate word by the query number qv and the click count elk.
  • the specific query process for determining the recommended query words related to the query word "Wang Fei” can be referred to the prior art, and will not be repeated herein.
  • the vertical search propensity recommendation module After determining the recommended query word list, the vertical search propensity recommendation module queries the vertical search propensity weight corresponding to each recommended query word and determines the propensity vertical search type of the recommended query word according to the preference weight.
  • the specific method for determining the preferred vertical search type of the recommended query word according to the preference weight may be two types, one is to use the result of the above-mentioned preference weight filtering, if the recommended query word is perpendicular to all types If the search bias weight is 0, then the recommended query is determined.
  • the word has no propensity vertical search type; otherwise, the vertical search type with the highest preference weight of the recommended query word is determined as the preferred vertical search type of the recommended query word; the other is the processing without filtering the above bias weight
  • the vertical search type with the highest preference weight of the recommended query word is directly determined as the preferred vertical search type of the recommended query word.
  • the recommendation display module After determining the recommended query words and their orientation vertical search types, the recommendation display module displays the links of the recommended query words and their preference vertical search types, as shown in FIG. 5, the display result is shown in FIG. 5. In the middle, the preference weight filtering process is used. If a recommended query word (such as the recommended query word "Wang Fei Weibo" in FIG. 5) has a weight of 0 for all types of vertical search, then it is determined. The recommended query word does not have a propensity vertical search type, and does not display links to its preferred vertical search type.
  • a recommended query word such as the recommended query word "Wang Fei Weibo" in FIG. 5
  • the recommended query word does not have a propensity vertical search type, and does not display links to its preferred vertical search type.
  • the response module detects that the user clicks on the link of the preferred vertical search type corresponding to a recommended query word
  • the vertical search engine corresponding to the vertical search type searches for the matching content of the recommended query word, and then jumps to the corresponding vertical Search type pages display search results.
  • the user can click the vertical search type link "@Video” corresponding to the query word "Wang Fei Xi'an Concert” in Figure 5, and then call the vertical search engine of the video class to search for the query word "Wang Fei Xi'an Concert" and will return The video search results shown in Figure 6.
  • the webpage search engine searches for the matching content of the recommended query word, and displays the webpage search result.
  • the display process of this webpage search result is the same as the prior art, and will not be described in detail herein.
  • the link of the vertical search type tends to be added, thereby providing the search user with a more intuitive and intelligent vertical search link, and guiding the user to directly reach the corresponding vertical search result. It improves the efficiency of vertical search and saves users' double click behavior, which not only reduces the number of human-computer interactions, but also saves computing and bandwidth resources, and provides users with more satisfactory search results, enabling users to obtain better search. Experience.
  • FIG. 4 is a schematic diagram showing the composition of a recommendation search device of a search engine according to an embodiment of the present invention.
  • the recommendation search device is used to perform the method of the embodiment of the present invention, which mainly includes:
  • the logging module 401 is configured to record a search log and a click log of the query words in the user history search process.
  • the historical search process refers to each search of the user within a specified time.
  • the process, in which "history" is relative to the query term recommendation module 403, refers to a search process before the user inputs the query word in real time in the search box.
  • the trend analysis module 402 is configured to analyze the vertical search tendency weight of the query word according to the recorded log.
  • the query word recommendation module 403 is configured to detect a query word in the search box, and determine a recommended query word related to the query word after detecting the query word.
  • the vertical search propensity recommendation module 404 is configured to query, from the preference analysis module 402, a vertical search propensity weight corresponding to each recommended query word, and determine a propensity vertical search type of the recommended query word according to the preference weight.
  • the recommendation display module 405 is configured to display the links of the recommended query words and their preferred vertical search types.
  • the response module 406 is configured to: after detecting that the link of the propensity vertical search type corresponding to the recommended query word is clicked, search for the matching content of the recommended query word by using a vertical search engine corresponding to the vertical search type, and display a search of the vertical search result.
  • the log recording module 401 may be specifically configured to: record, for each type of vertical search engine, each search in the vertical search engine The content of the entered query term, the number of links clicked in the search results, and the type of the vertical search engine;
  • the log recording module 401 may be specifically configured to: record the content of the query word and the search result that are input each time the search is performed in the webpage search engine.
  • the preference analysis module 402 is specifically configured to: determine a preference weight of each query word for each type of vertical search type.
  • the preference analysis module 402 is configured to: when targeting a certain query word to a certain vertical search type;
  • the first preference weight of the query word for the vertical search is analyzed according to the search log, including: determining a query weight X/Y, where X is a specified time period, and the query word is in the vertical search engine of the class.
  • the number of queries, Y is the total number of queries of the vertical search engine of the specified time period; determining the click weight N/M, where N is the query term within the specified time period.
  • the number of links clicked in the search result of the vertical search engine, M is the number of links clicked in the total search result of the vertical search engine page in the specified time period; weighting the query weight and the click weight
  • the first preference weight of the query word for the vertical search is analyzed according to the search log, including: determining a query weight X/Y, where X is a specified time period, and the query word is in the vertical search engine of the class.
  • Number of queries, Y is the total number of queries of the vertical search engine of the specified time period; determining the click weight N/M, where N is the search term of the vertical search engine in the specified time period.
  • the number of links clicked, M is the number of links clicked on the total search results of the vertical search engine pages in the specified time period; using the formula ( ⁇ ⁇ / ⁇ + ⁇ N/M ) to get the first tendency Sexual weight, where ⁇ is the query weight factor and ⁇ is the click weight factor; according to the click log, the second preference weight of the query word for the vertical search is analyzed, and the second bias weight is y/x, where X is The number of queries of the query word in the webpage search engine within a specified time period, and y is the number of links corresponding to the
  • the first preference weight of the query word for the vertical search is analyzed according to the search log, including: determining a query weight X/Y, where X is a specified time period, and the query word is in the vertical search engine of the class.
  • Number of queries, Y is the total number of queries of the vertical search engine of the specified time period;
  • determining the click weight N/M where N is the search term of the vertical search engine in the specified time period
  • the number of links clicked, M is the number of links clicked on the total search results of the vertical search engine pages in the specified time period; ⁇ using the formula ( 1+X/Y ) ( 1+N/M ) and returning a floating point number of the interval of (0,1), the floating point number being the first tendency weight; analyzing the second tendency weight of the query word for the vertical search according to the click log, the second The propensity weight is y/x, where X is the number of queries of the query word in the web search engine within a specified time period, and y is the clicked result of the web
  • the preference analysis module 402 is further configured to: after analyzing the vertical search propensity weight of the query word according to the recorded log, filtering the preference weight, specifically: The vertical search propensity weight is compared to the set threshold, and if the vertical search propensity weight of a certain class is lower than the threshold, the vertical search propensity weight is set to zero.
  • the vertical search propensity recommendation module 404 is specifically configured to: query a vertical search propensity weight corresponding to each recommended query word, and if the recommended query word has a bias weight of 0 for all types of vertical search, determine the recommendation The query word has no tendency vertical search type; otherwise, the vertical search type with the highest preference weight of the recommended query word is determined as the preferred vertical search type of the recommended query word.
  • the vertical search propensity recommendation module 404 is specifically configured to: query a vertical search propensity weight corresponding to each recommended query word, and determine a vertical search type with the highest preference weight of the recommended query word as the recommended query word. A preferred vertical search type.
  • the device of the embodiment adds a link of a vertical search type tendency by analyzing the tendency of the vertical search of the query word, provides a more intuitive and intelligent vertical search link for the search user, and guides the user to directly reach the corresponding vertical search.
  • the efficiency of vertical search is improved, and the behavior of secondary clicks is saved, which not only reduces the number of human-computer interactions, but also saves computational and bandwidth resources, and provides users with more satisfactory search results, so that users can get better. Search experience.
  • An embodiment of the present invention further provides a computer readable storage medium having stored thereon an instruction set, and when the instruction set is executed, performing the method described in any of the above embodiments.
  • the computer readable storage medium may be a floppy disk, a hard disk or an optical disk of a computer, and the like may be a mobile phone, a personal computer, a server, or a network device.

Abstract

本发明实施例公开了一种搜索引擎的推荐搜索方法、装置及计算机可读存储介质,包括:日志记录模块记录搜索过程中查询词的搜索日志和点击日志;倾向性分析模块根据日志分析查询词的垂直搜索倾向性权重;查询词推荐模块在搜索框检测到查询词后确定相关的推荐查询词;垂直搜索倾向性推荐模块查询各推荐查询词对应的垂直搜索倾向性权重并确定推荐查询词的倾向性垂直搜索类型;推荐显示模块显示所述各推荐查询词及其倾向性垂直搜索类型的链接;响应模块在检测到推荐查询词对应的倾向性垂直搜索类型的链接被点击后,利用该垂直搜索类型对应的垂直搜索引擎搜索该推荐查询词的匹配内容,展示搜索结果。利用本发明可以提高搜索引擎进行垂直搜索的搜索效率。

Description

搜索引擎的推荐搜索方法、 装置及计算机可读存储介质 技术领域 本发明涉及互联网信息搜索技术, 尤其涉及一种搜索引擎的推荐搜索 方法、 装置及计算机可读存储介质。 背景技术
互联网搜索引擎是指根据一定的策略、 运用特定的程序从互联网上搜集 信息, 在对信息进行组织和处理后, 为用户提供检索服务。 搜索引擎根据用 户输入的关键词检索出的相关信息, 并将检索出的相关信息作为搜索结果展 示给用户。
搜索引擎技术发展的一个重要目标就是尽量减少用户的人机操作次数, 尽量向用户展示符合用户搜索意图的搜索结果。 为了达到这个目标, 搜索引 擎技术正在不断地改进和发展。
目前, 出现了一种搜索引擎的推荐查询词展示技术, 这种技术的主要处 理过程是: 用户在搜索引擎的搜索检索框中输入查询词的过程中, 搜索引擎 会根据用户的输入的查询词,如查询词 abc,通过文本索引找出命中 abc的候 选查询词, 再通过查询次数和 /或点击率等统计数据对候选词进行筛选得到最 终推荐给用户的查询词 (在业界这种推荐的查询词也被称为推荐查询词) , 并实时显示所述最终推荐给用户的推荐查询词, 从而帮助用户筛选出可能感 兴趣的查询词, 节约用户输入最终查询词的时间, 提高搜索效率。 图 1为现 有技术中用户输入查询词时搜索引擎自动计算显示出推荐查询词的一种界面 图, 如图 1所示, 用户在搜索框中输入了查询词 "王菲" 101 , 则搜索引擎立 刻进行推荐计算处理得到推荐查询词, 并显示出包含所述推荐查询词的列表 102。 当用户对推荐出的某个推荐查询词感兴趣时, 则可能会点击该推荐查询 词, 完成一次网页搜索的行为。
上述这种方式虽然节约了用户输入的时间, 但是网页的搜索结果并不能 4艮好的满足用户的垂直搜索需求。 所谓垂直搜索, 就是针对某一行业的专业 搜索技术, 是搜索引擎的细分和延伸, 是对网页库中的某类专门的信息进行 一次整合, 定向分字段抽取出需要的数据进行处理后再以某种形式返回给用 户。 相对通用搜索引擎的信息量大、 查询不准确、 深度不够等提出来的新的 搜索引擎技术方式, 通过针对某一特定领域、 某一特定人群或某一特定需求 提供的有一定价值的信息和相关服务。 其特点就是 "专、 精、 深" , 且具有 行业色彩, 相比较通用搜索引擎的海量信息无序化, 垂直搜索引擎则显得更 力口专注、 具体和深入。
例如, 目前搜索引擎的网页搜索引擎为一种通用的、 综合的搜索技术, 其根据查询词搜索出的结果包含各种细分信息类型的结果, 例如其搜索结果 页面中嚢括了视频、 图片、 新闻、 音乐等各种类型的搜索结果。 而垂直搜索 需要将不同的信息类型区分开来,一种垂直搜索引擎只搜索一种类型的内容。 例如视频垂直搜索引擎专门用于搜索视频类的结果, 新闻垂直搜索引擎专门 用来搜索新闻类的搜索结果。 目前的大多数搜索引擎都具备了不同的垂直搜 索引擎(业界也称所述垂直搜索引擎为垂直搜索频道) , 分别对应搜索不同 种类的垂直搜索类型。 如图 2所示为现有一种搜索引擎主页的界面示意图, 其中包括网页搜索引擎 201 (即通用的搜索引擎),也包括图片搜索引擎 202、 视频搜索引擎 203、音乐搜索引擎 204、问答搜索引擎 205(即图中的 "问问 " )、 新闻搜索引擎 206等等这种垂直搜索引擎。
图 1所述的现有技术虽然节约了用户输入的时间, 但是并不能艮好的满 足用户的垂直搜索需求。比如图 1中如果用户点击了推荐查询词"王菲 传奇" 103或 "王菲西安演唱会" 104, 则会直接显示对应的网页搜索结果。 但是, 不同的推荐查询词的实际搜索意图大部分对应相应的垂直搜索类型,例如 "王 菲 传奇" 103的实际搜索意图大部分对应音乐内容, 而 "王菲西安演唱会" 104的实际搜索意图大部分对应视频内容。 而在图 1所示的现有技术中, 不 能单独列出更为直观的相关性较高的垂直搜索种类, 如果查询对应的垂直搜 索内容还需要用户进行二次点击相关的垂直搜索频道链接, 才能找到对应的 垂直搜索内容, 例如音乐类内容或者视频类内容。
因此现有技术在垂直搜索时的搜索效率不高, 不方便用户从搜索结果中 找到与查询词相关性较高的垂直搜索结果, 用户也无法得知与各个推荐查询 词相关性最高的垂直搜索类型; 同时, 用户为了选择最终的垂直搜索结果, 往往需要二次点击相关的垂直搜索频道链接, 导致用户与互联网机器侧的人 机交互次数增多, 而每一次人机交互操作都会发出操作请求信息、 触发计算 过程并产生响应结果信息,从而会占用机器侧的大量资源, 包括客户端资源、 服务器资源、 网络带宽资源等等。 发明内容
有鉴于此, 本发明实施例的主要目的在于提供一种搜索引擎的推荐搜 索方法、 装置及计算机可读存储介质, 以提高垂直搜索的搜索效率。
本发明实施例的技术方案是这样实现的:
一种搜索引擎的推荐搜索方法, 包括:
记录搜索过程中查询词的搜索日志和点击曰志;
根据所记录的日志分析查询词的垂直搜索倾向性权重;
在搜索框中检测到查询词之后, 确定与该查询词相关的推荐查询词, 查询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向性权重确 定推荐查询词的倾向性垂直搜索类型, 显示所述各推荐查询词及其倾向性 垂直搜索类型的链接;
在检测到推荐查询词对应的倾向性垂直搜索类型的链接被点击后, 利 用该垂直搜索类型对应的垂直搜索引擎搜索该推荐查询词, 展示搜索结 果。
一种搜索引擎的推荐搜索装置, 包括:
日志记录模块, 用于记录搜索过程中查询词的搜索日志和点击日志; 倾向性分析模块, 用于根据所记录的日志分析查询词的垂直搜索倾向 性权重;
查询词推荐模块, 用于在搜索框中检测查询词, 在检测到查询词后确 定与该查询词相关的推荐查询词;
垂直搜索倾向性推荐模块, 用于查询各推荐查询词对应的垂直搜索倾 向性权重并根据所述倾向性权重确定推荐查询词的倾向性垂直搜索类型; 推荐显示模块, 用于显示所述各推荐查询词及其倾向性垂直搜索类型 的链接;
响应模块, 用于在检测到推荐查询词对应的倾向性垂直搜索类型的链 接被点击后, 利用该垂直搜索类型对应的垂直搜索引擎搜索该推荐查询词 的匹配内容, 展示搜索结果。
一种计算机可读存储介质, 其上存储有指令集合, 当该指令集合被执 行时, 使得该计算机可执行上述任意一种搜索引擎的推荐搜索方法
与现有技术相比, 本发明实施例通过对查询词进行垂直搜索的倾向性 分析, 加入垂直搜索类型倾向性的链接, 为搜索用户提供更为直观的、 智 能化的垂直搜索链接, 引导用户直接到达相应的垂直搜索结果, 提高了垂 直搜索的效率, 同时还节省用户二次点击的行为, 不但降低了人机交互次 数, 节省了计算和带宽资源, 而且为用户提供更加满意的搜索结果, 使用 户获得更好的搜索体验。 附图说明 图 1为现有技术中用户输入查询词时搜索引擎自动计算显示出推荐查 询词的一种界面图;
图 2为现有一种搜索引擎主页的界面示意图;
图 3为本发明实施例所述一种搜索引擎的推荐搜索方法的处理流程图 图 4为本发明实施例所述一种搜索引擎的推荐搜索装置的组成示意 图;
图 5为本发明实施例在用户向搜索框中输入查询词时所述推荐搜索装 置自动分析显示出推荐查询词及其垂直搜索倾向性类型的一种界面图; 图 6为本发明实施例所述在用户点击推荐查询词对应的倾向性垂直搜 索类型链接后的搜索结果示意图;
图 7为在网页搜索结果中显示的视频类搜索结果的界面示意图。 具体实施方式
下面结合附图及具体实施例对本发明实施例再作进一步详细的说明。 图 3 为本发明实施例所述一种搜索引擎的推荐搜索方法的处理流程 图。 参见图 3 , 该方法主要包括:
步骤 301、 记录搜索过程中查询词的搜索日志和点击日志。
步骤 302、 根据所记录的日志分析查询词的垂直搜索倾向性权重。 步骤 303、 在搜索框中检测到查询词之后, 确定与该查询词相关的推 荐查询词, 查询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向 性权重确定推荐查询词的倾向性垂直搜索类型, 并显示所述各推荐查询词 及其倾向性垂直搜索类型的链接。
步骤 304、 在检测到推荐查询词对应的倾向性垂直搜索类型的链接被 点击后, 利用该垂直搜索类型对应的垂直搜索引擎搜索所述推荐查询词的 匹配内容, 展示搜索结果。
根据该方法, 通过对查询词进行垂直搜索的倾向性分析, 加入垂直搜 索类型倾向性的链接, 为搜索用户提供更为直观的、 智能化的垂直搜索链 接, 引导用户直接到达相应的垂直搜索结果, 提高了垂直搜索的效率, 同 时还节省用户二次点击的行为, 不但降低了人机交互次数, 节省了计算和 带宽资源, 而且为用户提供更加满意的搜索结果, 使用户获得更好的搜索 体验。 图 5为本发明实施例在用户向搜索框中输入查询词时所述推荐搜索装 置自动分析显示出推荐查询词及其垂直搜索倾向性类型的一种界面图。 如 图 5所示, 当用户在搜索引擎的输入框中输入查询词 501时, 本发明实施 例所述的推荐搜索装置会自动确定出与用户输入的查询词相关联的一组 查询词 502 , 并将该组查询词 502显示在所述搜索框下方, 并且对其中倾 向性明确的查询词给出对应的倾向性垂直搜索类型的链接 503显示在对应 查询词的旁边,例如图 5所示为显示在对应查询词后的符号 @之后。此后, 用户不仅可以点击其中的某个推荐查询词进行网页搜索, 而且可以直接点 击所述符号 @后面的倾向性垂直搜索类型链接直接调用对应类型的垂直 搜索引擎对相应的查询词进行搜索, 比如用户可点击图 5中查询词 "王菲 西安演唱会" 对应的垂直搜索类型链接 "@视频" , 则会调用视频类的垂 直搜索引擎搜索所述查询词 "王菲西安演唱会" , 并会返回如图 6所示的 视频搜索结果。 通过本发明实施例的方案, 可以节省用户在大量网页搜索 结果中查找感兴趣的垂直资源的过程, 得到更加符合用户需求的搜索结 果, 提高了搜索效率。
所述图 5 和图 6 所示的界面实例是以在计算机搜索引擎为例进行说 明, 在无线搜索场景下, 本发明实施例同样可以适用无线搜索引擎。 下面对本发明实施例所述各个步骤进行更为详细的说明: 步骤 301中所述记录搜索过程中查询词的搜索日志和点击日志是在用 户的搜索过程中进行的。 所述搜索过程是指在进行本发明实施例所述的推 荐查询词及其倾向性垂直搜索类型之前的搜索过程, 也可以称为相对的历 史搜索过程。 这种搜索过程主要包括: 用户在使用搜索引擎时提交一个搜 索请求, 其中包括查询词, 前端服务器得到该搜索请求后向搜索引擎的后 台系统发出检索请求, 搜索引擎经过检索后返回给用户一个搜索结果列 表。 如果用户找到感兴趣的搜索结果, 则会点击其中的某个或者某几个搜 索结果链接进行查看。 在这个搜索过程中, 本发明实施例的日志记录模块 需要记录两种日志: 一种是搜索日志, 一种是点击日志。
其中, 所述搜索日志主要是针对各类型的垂直搜索引擎, 记录查询词 的搜索日志的具体方法为: 针对各类垂直搜索引擎, 记录在该类垂直搜索 引擎中每次搜索时输入的查询词的内容、 搜索结果中被点击的链接数、 以 及该垂直搜索引擎的类型;
所述点击日志主要是针对网页搜索引擎, 记录查询词的点击日志的具 体方法为: 记录在网页搜索引擎中每次搜索时输入的查询词的内容、 搜索 结果中被点击的链接、 被点击链接对应的垂直搜索类型。
在得到所述搜索日志和点击日志之后, 则将这两种日志数据输入到倾 向性分析模块进行分析计算, 利用步骤 302的方法得到一个带有倾向性的 查询词推荐集合。
步骤 302所述根据所记录的日志分析查询词的垂直搜索倾向性权重具 体是确定各查询词对各类垂直搜索类型的倾向性权重, 其中针对某一查询 词对某一垂直搜索类型的倾向性权重的具体方法包括如下步骤 321 至 323 , 下面以查询词 "王菲西安演唱会" 对视频类垂直搜索的倾向性权重 为例进行说明:
步骤 321 , 根据所述搜索日志分析该查询词 "王菲西安演唱会" 对该 视频类垂直搜索的第一倾向性权重 pwl , 具体的确定方法包括: 确定查询 权重 X/Y, 其中 X为在指定时间段内, 该查询词在该视频类垂直搜索引擎 中的查询次数, Y为所述指定时间段内该视频类垂直搜索引擎的总查询次 数; 确定点击权重 N/M, 其中 N为所述指定时间段内该查询词 "王菲西安 演唱会"在该视频类垂直搜索引擎的搜索结果中被点击的链接数, M为所 述指定时间段内该视频类垂直搜索引擎页面的总搜索结果(即该视频类垂 直搜索引擎所有查询词的搜索结果) 中被点击的链接数; 对所述查询权重 和点击权重进行加权计算得到该查询词 "王菲西安演唱会" 对视频类垂直 搜索的第一倾向性权重 pw l。
具体的, 所述对查询权重和点击权重进行加权计算得到所述查询词对 所述垂直搜索类型的第一倾向性权重的具体方式可以使用加权求和的方 式得到, 也可以使用乘积形式得到。
所述利用加权求和的方式计算第一倾向性权重 pwl具体为: 釆用公式 ( α Χ/Υ + β Ν/Μ ) , 得到第一倾向性权重 pwl , 其中 α为查询权 重因子, β为点击权重因子, 这两个因子可以预先设定好。
所述利用乘积形式计算第一倾向性权重 pwl 具体为: 釆用公式 ( 1+X/Y ) ( 1 +N/M ) 并归一化到 (0, 1 ) 区间的一个浮点数, 该浮点数 为所述第一倾向性权重 pw l。
步骤 322、 根据所述点击日志分析该查询词 "王菲西安演唱会" 对视 频类垂直搜索的第二倾向性权重 pw2 , 该第二倾向性权重 pw2为 y/x , 其 中 X为该查询词在指定时间段内在网页搜索引擎中的查询次数, 例如在指 定时间段 t内用户在网页搜索引擎页面中搜索了 X次 "王菲西安演唱会"; y为该查询词的网页搜索结果中被点击的与视频类垂直搜索对应的链接的 个数, 例如用户点击了针对查询词 "王菲西安演唱会" 的搜索结果中的 y 条视频类的搜索结果。 图 7为在网页搜索结果中显示的视频类搜索结果的 界面示意图,比如用户点击了图 Ί中的 3条视频类搜索结果 701、702、703 , 则 y=3 ,如果用户点击的视频类搜索结果越多,则所述第二倾向性权重 pw2 越高, 表示用户对查询词 "王菲西安演唱会" 对应的视频类搜索结果感兴 趣, 该查询词的视频类垂直搜索的倾向性越高。
步骤 323、将所述第一倾向性权重 pwl和第二倾向性权重 pw2相乘得 到该查询词针对该垂直搜索类型的倾向性权重。
通过对各类垂直搜索引擎的搜索日志以及网页搜索引擎的点击曰志 的分析计算, 最终可以得到一组带有倾向性权重的推荐查询词, 如下表 1 : 查询词 查询次数 qv 点击次数 elk 倾向性权重 王菲的微博 303 230 0
王菲的歌曲 260 300 音乐 0.9
王菲西安演唱会 230 320 音乐 0.9 ,视频 0.8 王菲 传奇 200 260 音乐 0.9
王菲李亚鹏 131 100 新闻 0.7
王菲成名史 1 10 80 问问 0.5
王菲最新消息 103 123 新闻 0.6 表 1
本实施例中可以设定一个倾向性权重的阈值, 用于对所述倾向性权重 进行过滤, 如果某查询词的某类垂直搜索类型的倾向性权重低于该设定的 阈值, 在表示无明确倾向性, 则该查询词对应的该类垂直搜索类型的倾向 性权重为 0 , 如果该查询词对应的各类的垂直搜索类型的倾向性权重都为 0 , 则表 1中该查询词对应的倾向性权重可以设为 0。 如果某查询词的某类 垂直搜索类型的倾向性权重大于所述阈值, 则表示该查询词对该类垂直搜 索类型具有一定的倾向性, 其倾向性权重分值越高, 表示倾向性越高。 对 上述这些查询词建立索引, 由本发明实施例的推荐搜索装置加载到内存 中。
当检测到用户在搜索框中输入查询词 "王菲" 时, 本发明实施例的查 询词推荐模块首先通过文本索引找出命中王菲的候选词, 再通过查询次数 qv和点击次数 elk对候选词进行筛选得到最终推荐给用户的推荐查询词, 所述确定与该查询词 "王菲" 相关的推荐查询词的具体过程可以参见现有 技术, 本文不再赘述。
在确定了所述推荐查询词列表之后, 所述垂直搜索倾向性推荐模块查 询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向性权重确定 推荐查询词的倾向性垂直搜索类型。
所述根据所述倾向性权重确定推荐查询词的倾向性垂直搜索类型的 具体方法可以有两种, 一种是釆用上述倾向性权重过滤的结果, 如果所述 推荐查询词对所有类型的垂直搜索倾向性权重都为 0 , 则确定该推荐查询 词没有倾向性垂直搜索类型; 否则将所述推荐查询词倾向性权重最高的垂 直搜索类型确定为该推荐查询词的倾向性垂直搜索类型; 另一种是不釆用 上述倾向性权重过滤的处理, 直接将所述推荐查询词的倾向性权重最高的 垂直搜索类型确定为该推荐查询词的倾向性垂直搜索类型。
在确定各推荐查询词及其倾向性垂直搜索类型之后, 由所述推荐显示 模块显示所述各推荐查询词及其倾向性垂直搜索类型的链接, 如图 5所示 为显示结果, 该图 5中是釆用了所述倾向性权重过滤处理的, 如果某推荐 查询词 (如图 5中的推荐查询词 "王菲微博" 等)对所有类型的垂直搜索 倾向性权重都为 0 , 则确定该推荐查询词没有倾向性垂直搜索类型, 不显 示其倾向性垂直搜索类型的链接。
如果响应模块在检测到用户点击了某一推荐查询词对应的倾向性垂 直搜索类型的链接后, 利用该垂直搜索类型对应的垂直搜索引擎搜索该推 荐查询词的匹配内容, 然后跳到相应的垂直搜索类型的页面展示搜索结 果。 例如用户可点击图 5中查询词 "王菲西安演唱会" 对应的垂直搜索类 型链接 "@视频" , 则会调用视频类的垂直搜索引擎搜索所述查询词 "王 菲西安演唱会" , 并会返回如图 6所示的视频搜索结果。
如果所述响应模块检测到用户点击了某一推荐查询词, 而不是该推荐 查询词后面的倾向性垂直搜索类型链接, 则利用网页搜索引擎搜索该推荐 查询词的匹配内容, 展示网页搜索结果, 这个网页搜索结果的展示过程与 现有技术相同, 本文不再赘述。
根据该方法, 通过对查询词进行垂直搜索的倾向性分析, 加入垂直搜 索类型倾向性的链接, 为搜索用户提供更为直观的、 智能化的垂直搜索链 接, 引导用户直接到达相应的垂直搜索结果, 提高了垂直搜索的效率, 同 时还节省用户二次点击的行为, 不但降低了人机交互次数, 节省了计算和 带宽资源, 而且为用户提供更加满意的搜索结果, 使用户获得更好的搜索 体验。
图 4 为本发明实施例所述一种搜索引擎的推荐搜索装置的组成示意 图。 该推荐搜索装置用于执行本发明实施例的方法, 主要包括:
日志记录模块 401 , 用于记录用户历史搜索过程中查询词的搜索日志 和点击日志。 所述的历史搜索过程是指在指定的时间内的用户每一次搜索 过程, 其中的 "历史" 是相对于查询词推荐模块 403而言, 是指用户在搜 索框实时输入查询词之前的搜索过程。
倾向性分析模块 402, 用于根据所记录的日志分析查询词的垂直搜索 倾向性权重。
查询词推荐模块 403 , 用于在搜索框中检测查询词, 在检测到查询词 后确定与该查询词相关的推荐查询词。
垂直搜索倾向性推荐模块 404 , 用于从所述倾向性分析模块 402中查 询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向性权重确定 推荐查询词的倾向性垂直搜索类型。
推荐显示模块 405 , 用于显示所述各推荐查询词及其倾向性垂直搜索 类型的链接。
响应模块 406, 用于在检测到推荐查询词对应的倾向性垂直搜索类型 的链接被点击后, 利用该垂直搜索类型对应的垂直搜索引擎搜索所述推荐 查询词的匹配内容, 展示垂直搜索的搜索结果。
可选地, 日志记录模块 401在执行记录查询词的搜索日志的步骤时, 该曰志记录模块 401具体可以用于: 针对各类垂直搜索引擎, 记录在该类 垂直搜索引擎中每次搜索时输入的查询词的内容、 搜索结果中被点击的链 接数、 以及该垂直搜索引擎的类型;
可选地, 日志记录模块 401在执行记录查询词的点击日志的步骤时, 该日志记录模块 401具体可以用于: 记录在网页搜索引擎中每次搜索时输 入的查询词的内容、 搜索结果中被点击的链接、 被点击链接对应的垂直搜 索类型。
可选地, 倾向性分析模块 402具体用于: 确定各查询词对各类垂直搜 索类型的倾向性权重。
可选地, 倾向性分析模块 402在针对某一查询词对某一垂直搜索类型 的倾向性权重时, 该倾向性分析模块 402具体用于:
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 类垂直搜索引擎页面的总搜索结果中被点击的链接数; 对所述查询权重和 点击权重进行加权计算得到该查询词对该类垂直搜索的第一倾向性权重; 根据所述点击日志分析该查询词对该类垂直搜索的第二倾向性权重, 该第 二倾向性权重为 y/x , 其中 X为该查询词在指定时间段内在网页搜索引擎 中的查询次数, y为该查询词的网页搜索结果中被点击的与该类垂直搜索 对应的链接的个数; 将所述第一倾向性权重和第二倾向性权重相乘得到该 查询词针对该垂直搜索类型的倾向性权重; 或者
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 类垂直搜索引擎页面的总搜索结果中被点击的链接数;釆用公式( χ Χ/Υ + β N/M ) , 得到第一倾向性权重, 其中 α为查询权重因子, β为点 击权重因子; 根据所述点击日志分析该查询词对该类垂直搜索的第二倾向 性权重, 该第二倾向性权重为 y/x其中 X为该查询词在指定时间段内在网 页搜索引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与该 类垂直搜索对应的链接的个数; 将所述第一倾向性权重和第二倾向性权重 相乘得到该查询词针对该垂直搜索类型的倾向性权重; 或者
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 类垂直搜索引擎页面的总搜索结果中被点击的链接数; 釆用公式( 1+X/Y ) ( 1+N/M )并归一化到 (0,1 ) 区间的一个浮点数, 该浮点数为所述第一 倾向性权重; 根据所述点击日志分析该查询词对该类垂直搜索的第二倾向 性权重, 该第二倾向性权重为 y/x , 其中 X为该查询词在指定时间段内在 网页搜索引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与 该类垂直搜索对应的链接的个数; 将所述第一倾向性权重和第二倾向性权 重相乘得到该查询词针对该垂直搜索类型的倾向性权重。
可选地, 所述倾向性分析模块 402进一步用于: 在根据所记录的日志 分析查询词的垂直搜索倾向性权重之后, 对所述倾向性权重进行过滤, 具 体为: 将查询词的各类垂直搜索倾向性权重与设定的阈值进行比较, 如果 某类的垂直搜索倾向性权重低于该阈值则将该类垂直搜索倾向性权重设 置为 0。 此时, 垂直搜索倾向性推荐模块 404具体用于: 查询各推荐查询 词对应的垂直搜索倾向性权重, 如果所述推荐查询词对所有类型的垂直搜 索倾向性权重都为 0 , 则确定该推荐查询词没有倾向性垂直搜索类型; 否 则将所述推荐查询词倾向性权重最高的垂直搜索类型确定为该推荐查询 词的倾向性垂直搜索类型。
可选地, 垂直搜索倾向性推荐模块 404具体用于: 查询各推荐查询词 对应的垂直搜索倾向性权重, 将所述推荐查询词的倾向性权重最高的垂直 搜索类型确定为该推荐查询词的倾向性垂直搜索类型。
本实施例的装置通过对查询词进行垂直搜索的倾向性分析, 加入垂直 搜索类型倾向性的链接, 为搜索用户提供更为直观的、 智能化的垂直搜索 链接, 引导用户直接到达相应的垂直搜索结果, 提高了垂直搜索的效率, 同时还节省用户二次点击的行为, 不但降低了人机交互次数, 节省了计算 和带宽资源, 而且为用户提供更加满意的搜索结果, 使用户获得更好的搜 索体验。
本发明实施例还提供一种计算机可读存储介质, 其上存储有指令集 合, 当该指令集合被执行时, 执行上述任一实施例所述的方法。 该计算机 可读存储介质可以是计算机的软盘、硬盘或光盘等,该计算机可以是手机、 个人计算机、 服务器或者网络设备等。
以上所述仅为本发明的较佳实施例而已, 并不用以限制本发明, 凡在 本发明的精神和原则之内, 所做的任何修改、 等同替换、 改进等, 均应包 含在本发明保护的范围之内。

Claims

权 利 要 求 书
1、 一种搜索引擎的推荐搜索方法, 其特征在于, 包括:
记录搜索过程中查询词的搜索日志和点击曰志;
根据所记录的日志分析查询词的垂直搜索倾向性权重;
在搜索框中检测到查询词之后, 确定与该查询词相关的推荐查询词, 查询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向性权重确 定推荐查询词的倾向性垂直搜索类型, 显示所述各推荐查询词及其倾向性 垂直搜索类型的链接;
在检测到所述推荐查询词对应的倾向性垂直搜索类型的链接被点击 后, 利用该倾向性垂直搜索类型对应的垂直搜索引擎搜索所述推荐查询 词, 展示搜索结果。
2、 根据权利要求 1所述的方法, 其特征在于,
所述记录查询词的搜索日志的具体方法为: 针对各类垂直搜索引擎, 记录在该类垂直搜索引擎中每次搜索时输入的查询词的内容、 搜索结果中 被点击的链接数、 以及该垂直搜索引擎的类型;
所述记录查询词的点击日志的具体方法为: 记录在网页搜索引擎中每 次搜索时输入的查询词的内容、 搜索结果中被点击的链接、 被点击链接对 应的垂直搜索类型。
3、 根据权利要求 2所述的方法, 其特征在于, 所述根据所记录的日 志分析查询词的垂直搜索倾向性权重具体是确定各查询词对各类垂直搜 索类型的倾向性权重, 其中针对某一查询词对某一垂直搜索类型的倾向性 权重的具体方法为:
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 类垂直搜索引擎页面的总搜索结果中被点击的链接数; 对所述查询权重和 点击权重进行加权计算得到该查询词对该类垂直搜索的第一倾向性权重; 根据所述点击日志分析该查询词对该类垂直搜索的第二倾向性权重, 该第二倾向性权重为 y/χ , 其中 X为该查询词在指定时间段内在网页搜索 引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与该类垂直 搜索对应的链接的个数;
将所述第一倾向性权重和第二倾向性权重相乘得到该查询词针对该 垂直搜索类型的倾向性权重。
4、 根据权利要求 3 所述的方法, 其特征在于, 所述对查询权重和点 击权重进行加权计算得到所述查询词对所述垂直搜索类型的第一倾向性 权重的具体方式为: 釆用公式 ( α Χ/Υ + β Ν/Μ ) , 得到第一倾向 性权重, 其中 α为查询权重因子, β为点击权重因子。
5、 根据权利要求 3 所述的方法, 其特征在于, 所述对查询权重和点 击权重进行加权计算得到所述查询词对所述垂直搜索类型的第一倾向性 权重的具体方式为: 釆用公式 ( 1+Χ/Υ ) ( 1+N/M ) 并归一化到 (0,1 ) 区间的一个浮点数, 该浮点数为所述第一倾向性权重。
6、 根据权利要求 1所述的方法, 其特征在于,
所述根据所记录的日志分析查询词的垂直搜索倾向性权重之后, 进一 步包括: 对所述倾向性权重进行过滤, 具体为: 将查询词的各类垂直搜索 倾向性权重与设定的阈值进行比较, 如果某类的垂直搜索倾向性权重低于 该阈值则将该类垂直搜索倾向性权重设置为 0;
所述根据所述倾向性权重确定推荐查询词的倾向性垂直搜索类型的 具体方法为: 如果所述推荐查询词对所有类型的垂直搜索倾向性权重都为 0 , 则确定该推荐查询词没有倾向性垂直搜索类型; 否则将所述推荐查询 词倾向性权重最高的垂直搜索类型确定为该推荐查询词的倾向性垂直搜 索类型。
7、 根据权利要求 1所述的方法, 其特征在于,
所述根据所述倾向性权重确定推荐查询词的倾向性垂直搜索类型的 具体方法为: 将所述推荐查询词的倾向性权重最高的垂直搜索类型确定为 该推荐查询词的倾向性垂直搜索类型。
8、 根据权利要求 1 所述的方法, 其特征在于, 在检测到某一推荐查 询词被点击后, 利用网页搜索引擎搜索该推荐查询词的匹配内容, 展示搜 索结果。
9、 一种搜索引擎的推荐搜索装置, 其特征在于, 包括: 日志记录模块, 用于记录搜索过程中查询词的搜索日志和点击日志; 倾向性分析模块, 用于根据所记录的日志分析查询词的垂直搜索倾向 性权重;
查询词推荐模块, 用于在搜索框中检测查询词, 在检测到查询词后确 定与该查询词相关的推荐查询词;
垂直搜索倾向性推荐模块, 用于查询各推荐查询词对应的垂直搜索倾 向性权重并根据所述倾向性权重确定推荐查询词的倾向性垂直搜索类型; 推荐显示模块, 用于显示所述各推荐查询词及其倾向性垂直搜索类型 的链接;
响应模块, 用于在检测到推荐查询词对应的倾向性垂直搜索类型的链 接被点击后, 利用该垂直搜索类型对应的垂直搜索引擎搜索所述推荐查询 词的匹配内容, 展示搜索结果。
10、 根据权利要求 9所述的装置, 其特征在于, 所述响应模块进一步 用于在检测到某一推荐查询词被点击后, 利用网页搜索引擎搜索该推荐查 询词的匹配内容, 展示搜索结果。
11、 根据权利要求 9所述的装置, 其特征在于, 所述日志记录模块具 体用于:
针对各类垂直搜索引擎, 记录在该类垂直搜索引擎中每次搜索时输入 的查询词的内容、 搜索结果中被点击的链接数、 以及该垂直搜索引擎的类 型;
记录在网页搜索引擎中每次搜索时输入的查询词的内容、 搜索结果中 被点击的链接、 被点击链接对应的垂直搜索类型。
12、 根据权利要求 9所述的装置, 其特征在于, 所述倾向性分析模块 具体用于确定各查询词对各类垂直搜索类型的倾向性权重。
13、 根据权利要求 12所述的装置, 其特征在于, 所述倾向性分析模 块具体用于:
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 类垂直搜索引擎页面的总搜索结果中被点击的链接数; 对所述查询权重和 点击权重进行加权计算得到该查询词对该类垂直搜索的第一倾向性权重; 根据所述点击日志分析该查询词对该类垂直搜索的第二倾向性权重, 该第二倾向性权重为 y/χ , 其中 X为该查询词在指定时间段内在网页搜索 引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与该类垂直 搜索对应的链接的个数;
将所述第一倾向性权重和第二倾向性权重相乘得到该查询词针对该 垂直搜索类型的倾向性权重。
14、 根据权利要求 12所述的装置, 其特征在于, 所述倾向性分析模 块具体用于:
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 垂直搜索引擎页面的总搜索结果中被点击的链接数;釆用公式( χ Χ/Υ + β N/M ) , 得到第一倾向性权重, 其中 α为查询权重因子, β为点击 权重因子;
根据所述点击日志分析该查询词对该类垂直搜索的第二倾向性权重, 该第二倾向性权重为 y/χ , 其中 X为该查询词在指定时间段内在网页搜索 引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与该类垂直 搜索对应的链接的个数;
将所述第一倾向性权重和第二倾向性权重相乘得到该查询词针对该 垂直搜索类型的倾向性权重。
15、 根据权利要求 12所述的装置, 其特征在于, 所述倾向性分析模 块具体用于:
根据所述搜索日志分析该查询词对该类垂直搜索的第一倾向性权重, 包括: 确定查询权重 X/Y, 其中 X为在指定时间段内, 该查询词在该类垂 直搜索引擎中的查询次数, Y为所述指定时间段内该类垂直搜索引擎的总 查询次数; 确定点击权重 N/M,其中 N为所述指定时间段内该查询词在该 类垂直搜索引擎的搜索结果中被点击的链接数, M为所述指定时间段内该 垂直搜索引擎页面的总搜索结果中被点击的链接数; 釆用公式 ( 1+X/Y ) ( 1+N/M )并归一化到 (0,1 ) 区间的一个浮点数, 该浮点数为所述第一 倾向性权重;
根据所述点击日志分析该查询词对该类垂直搜索的第二倾向性权重, 该第二倾向性权重为 y/χ , 其中 X为该查询词在指定时间段内在网页搜索 引擎中的查询次数, y为该查询词的网页搜索结果中被点击的与该类垂直 搜索对应的链接的个数;
将所述第一倾向性权重和第二倾向性权重相乘得到该查询词针对该 垂直搜索类型的倾向性权重。
16、 根据权利要求 12所述的装置, 其特征在于, 倾向性分析模块进 一步用于:
在根据所记录的日志分析查询词的垂直搜索倾向性权重之后, 对所述 倾向性权重进行过滤, 具体为: 将查询词的各类垂直搜索倾向性权重与设 定的阈值进行比较, 如果某类的垂直搜索倾向性权重低于该阈值则将该类 垂直搜索倾向性权重设置为 0。
17、 根据权利要求 16所述的装置, 其特征在于, 所述垂直搜索倾向 性推荐模块具体用于:
查询各推荐查询词对应的垂直搜索倾向性权重, 如果所述推荐查询词 对所有类型的垂直搜索倾向性权重都为 0, 则确定该推荐查询词没有倾向 性垂直搜索类型; 否则将所述推荐查询词倾向性权重最高的垂直搜索类型 确定为该推荐查询词的倾向性垂直搜索类型。
18、 根据权利要求 9所述的装置, 其特征在于, 垂直搜索倾向性推荐 模块具体用于:
查询各推荐查询词对应的垂直搜索倾向性权重, 将所述推荐查询词的 倾向性权重最高的垂直搜索类型确定为该推荐查询词的倾向性垂直搜索 类型。
19、 一种计算机可读存储介质, 其上存储有指令集合, 当该指令集合 被执行时, 可执行如下步骤:
记录搜索过程中查询词的搜索日志和点击曰志;
根据所记录的日志分析查询词的垂直搜索倾向性权重;
在搜索框中检测到查询词之后, 确定与该查询词相关的推荐查询词, 查询各推荐查询词对应的垂直搜索倾向性权重并根据所述倾向性权重确 定推荐查询词的倾向性垂直搜索类型, 显示所述各推荐查询词及其倾向性 垂直搜索类型的链接;
在检测到所述推荐查询词对应的倾向性垂直搜索类型的链接被点击 后, 利用该倾向性垂直搜索类型对应的垂直搜索引擎搜索所述推荐查询 词, 展示搜索结果。
PCT/CN2013/073129 2012-03-26 2013-03-25 搜索引擎的推荐搜索方法、装置及计算机可读存储介质 WO2013143429A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/497,993 US9934312B2 (en) 2012-03-26 2014-09-26 Recommendation search method for search engine, device and computer readable storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201210081897.6A CN103365839B (zh) 2012-03-26 2012-03-26 一种搜索引擎的推荐搜索方法和装置
CN201210081897.6 2012-03-26

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/497,993 Continuation US9934312B2 (en) 2012-03-26 2014-09-26 Recommendation search method for search engine, device and computer readable storage medium

Publications (1)

Publication Number Publication Date
WO2013143429A1 true WO2013143429A1 (zh) 2013-10-03

Family

ID=49258208

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/073129 WO2013143429A1 (zh) 2012-03-26 2013-03-25 搜索引擎的推荐搜索方法、装置及计算机可读存储介质

Country Status (3)

Country Link
US (1) US9934312B2 (zh)
CN (1) CN103365839B (zh)
WO (1) WO2013143429A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109190049A (zh) * 2018-11-02 2019-01-11 北京字节跳动网络技术有限公司 关键词推荐方法、系统、电子设备和计算机可读介质

Families Citing this family (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103761326A (zh) * 2014-01-29 2014-04-30 百度在线网络技术(北京)有限公司 图片搜索方法以及搜索引擎
US9703859B2 (en) * 2014-08-27 2017-07-11 Facebook, Inc. Keyword search queries on online social networks
CN104268268B (zh) * 2014-10-13 2018-05-22 宁波公众信息产业有限公司 一种网页信息关联方法及系统
CN104572853B (zh) * 2014-12-17 2019-01-25 百度在线网络技术(北京)有限公司 搜索方法和搜索装置
CN104699751A (zh) * 2014-12-30 2015-06-10 北京奇虎科技有限公司 一种基于搜索词进行搜索推荐的方法和装置
CN104615723B (zh) * 2015-02-06 2018-08-07 百度在线网络技术(北京)有限公司 查询词权重值的确定方法和装置
CN105045901B (zh) * 2015-08-05 2019-04-30 百度在线网络技术(北京)有限公司 搜索关键词的推送方法和装置
CN105573887B (zh) * 2015-12-14 2018-07-13 合一网络技术(北京)有限公司 搜索引擎的质量评估方法和装置
CN105488221A (zh) * 2015-12-25 2016-04-13 北京奇虎科技有限公司 一种在搜索界面推荐查询词进行搜索的方法和系统
CN106250369B (zh) * 2016-07-28 2019-08-06 海信集团有限公司 语音交互方法、装置和终端
US20180089309A1 (en) * 2016-09-28 2018-03-29 Linkedln Corporation Term set expansion using textual segments
CN106503258B (zh) * 2016-11-18 2020-05-12 深圳市世强元件网络有限公司 一种网站站内精确搜索方法
CN108268522B (zh) * 2016-12-30 2021-01-22 北京国双科技有限公司 网站栏目内容的展现方法及装置
CN107577707B (zh) * 2017-07-31 2021-06-18 北京奇艺世纪科技有限公司 一种目标数据集生成方法、装置及电子设备
RU2692045C1 (ru) 2018-05-18 2019-06-19 Общество С Ограниченной Ответственностью "Яндекс" Способ и система для рекомендации свежих саджестов поисковых запросов в поисковой системе
KR102532203B1 (ko) * 2018-07-31 2023-05-12 삼성전자 주식회사 전압 생성기, 전압 파형 생성기, 반도체 소자 제조 장치, 전압 파형 발생 방법 및 반도체 소자의 제조 방법
TWI681304B (zh) * 2018-12-14 2020-01-01 財團法人工業技術研究院 自適應性調整關連搜尋詞的系統及其方法
CN110188261A (zh) * 2019-05-30 2019-08-30 江苏神州信源系统工程有限公司 一种热点对象确定方法及装置
CN110390052B (zh) * 2019-07-25 2022-10-28 腾讯科技(深圳)有限公司 搜索推荐方法、ctr预估模型的训练方法、装置及设备
WO2021227059A1 (zh) * 2020-05-15 2021-11-18 深圳市世强元件网络有限公司 一种基于多叉树的搜索词推荐方法及系统
CN111752436A (zh) * 2020-06-28 2020-10-09 北京搜狗科技发展有限公司 一种推荐方法、装置和用于推荐的装置
US20230214431A1 (en) * 2022-01-06 2023-07-06 Ebay Inc. Query modality recommendation for e-commerce search
CN115878905B (zh) * 2023-03-03 2023-05-05 北京缔业科技有限公司 一种智慧商圈的个性服务推荐系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100191740A1 (en) * 2009-01-26 2010-07-29 Yahoo! Inc. System and method for ranking web searches with quantified semantic features
CN102033877A (zh) * 2009-09-27 2011-04-27 阿里巴巴集团控股有限公司 检索方法和装置
CN102184230A (zh) * 2011-05-11 2011-09-14 北京百度网讯科技有限公司 一种搜索结果的展示方法及装置

Family Cites Families (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7428529B2 (en) * 2004-04-15 2008-09-23 Microsoft Corporation Term suggestion for multi-sense query
US8122019B2 (en) * 2006-02-17 2012-02-21 Google Inc. Sharing user distributed search results
KR100886767B1 (ko) * 2006-12-29 2009-03-04 엔에이치엔(주) 지유아이 검색 서비스 제공 방법 및 시스템
US8930356B2 (en) * 2007-09-20 2015-01-06 Yahoo! Inc. Techniques for modifying a query based on query associations
US8583670B2 (en) * 2007-10-04 2013-11-12 Microsoft Corporation Query suggestions for no result web searches
US8572112B2 (en) * 2007-11-02 2013-10-29 Microsoft Corporation Syndicating search queries using web advertising
US20100082649A1 (en) * 2008-09-22 2010-04-01 Microsoft Corporation Automatic search suggestions from server-side user history
US9317602B2 (en) * 2008-12-11 2016-04-19 Yahoo! Inc. System and method for in-context exploration of search results
WO2010076897A1 (en) * 2008-12-29 2010-07-08 Julien Yuki Hamonic A method for document retrieval based on queries that are composed of concepts and recommended terms
US8738641B2 (en) * 2009-03-31 2014-05-27 Ebay Inc. Ranking algorithm for search box auto-complete
IT1399261B1 (it) * 2009-06-11 2013-04-11 Galileo Avionica S P A Ora Selex Galileo Spa Rivelazione attiva a distanza di sostanze chimiche
AU2010286374B2 (en) * 2009-08-31 2015-09-17 Google Llc Framework for selecting and presenting answer boxes relevant to user input as query suggestions
US8676828B1 (en) * 2009-11-04 2014-03-18 Google Inc. Selecting and presenting content relevant to user input
US8504437B1 (en) * 2009-11-04 2013-08-06 Google Inc. Dynamically selecting and presenting content relevant to user input
US20110314419A1 (en) * 2010-06-22 2011-12-22 Microsoft Corporation Customizing a search experience using images
US8650173B2 (en) * 2010-06-23 2014-02-11 Microsoft Corporation Placement of search results using user intent
US8600979B2 (en) * 2010-06-28 2013-12-03 Yahoo! Inc. Infinite browse
CN102043833B (zh) * 2010-11-25 2013-12-25 北京搜狗科技发展有限公司 一种基于查询词进行搜索的方法和搜索装置
US9031970B1 (en) * 2011-01-26 2015-05-12 Google Inc. Query autocompletions
CN102360358B (zh) * 2011-09-28 2016-08-17 百度在线网络技术(北京)有限公司 关键词推荐方法及系统
CN102375885A (zh) * 2011-10-21 2012-03-14 北京百度网讯科技有限公司 一种提供与查询序列相对应的搜索建议的方法与设备
US9317605B1 (en) * 2012-03-21 2016-04-19 Google Inc. Presenting forked auto-completions

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100191740A1 (en) * 2009-01-26 2010-07-29 Yahoo! Inc. System and method for ranking web searches with quantified semantic features
CN102033877A (zh) * 2009-09-27 2011-04-27 阿里巴巴集团控股有限公司 检索方法和装置
CN102184230A (zh) * 2011-05-11 2011-09-14 北京百度网讯科技有限公司 一种搜索结果的展示方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109190049A (zh) * 2018-11-02 2019-01-11 北京字节跳动网络技术有限公司 关键词推荐方法、系统、电子设备和计算机可读介质
CN109190049B (zh) * 2018-11-02 2021-11-23 北京字节跳动网络技术有限公司 关键词推荐方法、系统、电子设备和计算机可读介质

Also Published As

Publication number Publication date
CN103365839B (zh) 2017-12-12
CN103365839A (zh) 2013-10-23
US20150012519A1 (en) 2015-01-08
US9934312B2 (en) 2018-04-03

Similar Documents

Publication Publication Date Title
WO2013143429A1 (zh) 搜索引擎的推荐搜索方法、装置及计算机可读存储介质
US11782970B2 (en) Query categorization based on image results
US7962477B2 (en) Blending mobile search results
US9317550B2 (en) Query expansion
US8751470B1 (en) Context sensitive ranking
US9336318B2 (en) Rich content for query answers
TWI582619B (zh) Method and apparatus for providing referral words
US20110314011A1 (en) Automatically generating training data
RU2670494C2 (ru) Способ обработки поискового запроса, сервер и машиночитаемый носитель для его осуществления
US8745081B2 (en) Personalization of news articles based on news sources
US20130212120A1 (en) Multi-domain recommendations
US10055755B2 (en) Targeting content based on receipt of partial terms
US10152521B2 (en) Resource recommendations for a displayed resource
US9542475B2 (en) Presenting content in accordance with a placement designation
US8612409B2 (en) Method and apparatus for detecting and explaining bursty stream events in targeted groups
US20200293787A1 (en) Media contextual information for a displayed resource
JP5787717B2 (ja) 情報処理装置、プログラム及び情報検索システム
US10146849B2 (en) Triggering answer boxes
US9424342B1 (en) Geographically local query detection
TWI483129B (zh) Retrieval method and device
US8161065B2 (en) Facilitating advertisement selection using advertisable units
JP2013092823A (ja) 情報処理装置、プログラム及び情報検索システム
Wu et al. A quality analysis of keyword searching in different search engines projects
US20210232659A1 (en) Query recommendations for a displayed resource
Laporte et al. Multiple Clicks Model for Web Search of Multi-clickable Documents

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13768177

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 13768177

Country of ref document: EP

Kind code of ref document: A1