WO2018000557A1 - 搜索结果展示方法和装置 - Google Patents

搜索结果展示方法和装置 Download PDF

Info

Publication number
WO2018000557A1
WO2018000557A1 PCT/CN2016/096480 CN2016096480W WO2018000557A1 WO 2018000557 A1 WO2018000557 A1 WO 2018000557A1 CN 2016096480 W CN2016096480 W CN 2016096480W WO 2018000557 A1 WO2018000557 A1 WO 2018000557A1
Authority
WO
WIPO (PCT)
Prior art keywords
core
webpage
search request
search
word segmentation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2016/096480
Other languages
English (en)
French (fr)
Inventor
孙珂
孙叔琦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and Technology Co Ltd
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 Beijing Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to US16/313,206 priority Critical patent/US10977317B2/en
Publication of WO2018000557A1 publication Critical patent/WO2018000557A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/906—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/957—Browsing optimisation, e.g. caching or content distillation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/31—Indexing; Data structures therefor; Storage structures
    • G06F16/313—Selection or weighting of terms for indexing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/3331—Query processing
    • G06F16/334—Query execution
    • G06F16/3344—Query execution using natural language analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/951—Indexing; Web crawling techniques
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9538—Presentation of query results

Definitions

  • the present invention relates to the field of information processing technologies, and in particular, to a search result display method and apparatus.
  • search engines as the main way to retrieve network information. For example, users can search for "Lingdu F8" in search engines to obtain search results related to "Lingdu F8". .
  • the search engine after inputting “Ling F8” in the search engine, the search result obtained by the search engine according to “Ling F8” is concentrated on the evaluation of “Ling F8”. Therefore, the search engine only provides users with information such as evaluation and price of "Lingdu F8", and the search results provided by the search engine are not sufficiently diversified and the repetition rate is high.
  • the object of the present invention is to solve at least one of the above technical problems to some extent.
  • the first object of the present invention is to provide a search result display method, which performs cluster analysis on a web page that matches a search request, and provides a user with a richer and more diverse search result.
  • a second object of the present invention is to provide a search result display method apparatus.
  • a third object of the invention is to propose an apparatus.
  • a fourth object of the present invention is to provide a non-volatile computer storage medium.
  • the first aspect of the present invention provides a search result display method, which acquires a plurality of web pages according to a user's search request, performs a special name recognition word segmentation process on each web page title, and obtains a corresponding word segmentation result.
  • a search result display method which acquires a plurality of web pages according to a user's search request, performs a special name recognition word segmentation process on each web page title, and obtains a corresponding word segmentation result.
  • the webpage matching the search request is subjected to cluster analysis, and the search result is generated and displayed to the user.
  • the search result display method of the embodiment of the present invention acquires multiple webpages according to a user's search request, for each webpage
  • the title is processed by a specific name recognition word segmentation, and the part of speech and extended information corresponding to the word segmentation result are obtained, and the core information of each web page is determined according to the part of speech and the extended information corresponding to the word segmentation result, and the result of the plurality of web pages is obtained according to the core information of each web page.
  • the webpage matching the search request is filtered, and the webpage matching the search request is clustered and analyzed, and the search result is displayed and displayed to the user. Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • search result display method of the embodiment of the present invention further has the following additional technical features:
  • the part of speech includes: a noun, a verb, an adverb, or a proper noun;
  • the extended information includes: a proper name category, a special name chain index analysis, or a vocabulary domain analysis.
  • the core information includes: a core proper noun and a core verb;
  • the filtering according to the core information of each webpage, the webpages that match the search request from the plurality of webpage results, including:
  • the webpage content includes the core proper noun and the core verb, detecting whether the TF-IDF of the core verb is higher than a preset threshold;
  • the core proper noun is consistent with the search request, it is determined to be a web page that matches the search request.
  • the core information further includes: a core concept word
  • Performing cluster analysis on the webpage that matches the search request including:
  • the clustering metric function is used to analyze the combined similarity of the core concept words and the core verbs in the participating clustering webpage;
  • the participating cluster pages are clustered.
  • the generating search results are displayed to the user, including:
  • the core concept words and the core verbs are presented to the user as the search result title.
  • An embodiment of the second aspect of the present invention provides a search result display apparatus, including:
  • a first obtaining module configured to acquire a plurality of webpages according to a search request of the user, perform a special name recognition word segmentation process on the title of each webpage, and obtain part of speech and extended information corresponding to the word segmentation result;
  • a second obtaining module configured to determine core information of each webpage according to the part of speech and extended information corresponding to the word segmentation result
  • a screening module configured to filter, according to core information of each webpage, a webpage that matches the search request from the plurality of webpage results
  • the cluster generation module is configured to perform cluster analysis on the webpage matching the search request, and generate a search result to display to the user.
  • the search result display device of the embodiment of the present invention acquires a plurality of web pages according to a search request of the user, performs a special name recognition word segmentation process on the title of each web page, and acquires part of speech and extended information corresponding to the word segmentation result, according to the word segmentation result corresponding to the word segmentation result.
  • the part of speech and the extended information determine the core information of each webpage, and the webpage matching the search request is filtered from the plurality of webpage results according to the core information of each webpage, and the webpage matching the search request is clustered and analyzed, and the search result is displayed and displayed. user. Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • search result display device of the embodiment of the present invention further has the following additional technical features:
  • the part of speech includes: a noun, a verb, an adverb, or a proper noun;
  • the extended information includes: a proper name category, a special name chain index analysis, or a vocabulary domain analysis.
  • the core information includes: a core proper noun and a core verb;
  • the screening module includes:
  • a first detecting unit configured to detect, according to core information of each webpage, whether the corresponding webpage content includes the core proper noun and the core verb;
  • a second detecting unit configured to detect, when the webpage content includes the core proper noun and the core verb, whether the TF-IDF of the core verb is higher than a preset threshold
  • a third detecting unit configured to detect, when the TF-IDF of the core verb is higher than a preset threshold, whether the core proper noun is consistent with the search request;
  • a determining unit configured to determine, when the core proper noun is consistent with the search request, a webpage that matches the search request.
  • the core information further includes: a core concept word
  • the cluster generation module includes:
  • a fourth detecting unit configured to detect whether the core proper nouns in the participating webpages are consistent
  • An analyzing unit configured to analyze, by using a clustering metric function, a combined similarity of the core concept words and the core verbs in the clustering webpage when the core proper nouns in the clustering webpage are consistent;
  • the clustering unit is configured to cluster the participating cluster pages when the combined similarity is greater than a preset threshold.
  • the cluster generation module further includes:
  • a display unit configured to display the core concept word and the core verb as the search result title to a user.
  • a third aspect of the present invention provides an apparatus comprising: one or more processors; a memory; one or more programs, the one or more programs being stored in the memory when When multiple processors are executed, the following steps are performed: acquiring multiple web pages according to a user's search request, and marking each web page The problem is to perform a special name recognition word segmentation process to obtain the part of speech and extended information corresponding to the word segmentation result;
  • the webpage matching the search request is subjected to cluster analysis, and the search result is generated and displayed to the user.
  • the device of the embodiment of the present invention acquires a plurality of web pages according to a search request of the user, performs a special name recognition word segmentation process on the title of each web page, acquires part of speech and extended information corresponding to the word segmentation result, and uses part of speech and extension corresponding to the word segmentation result.
  • the information determines the core information of each webpage, and filters the webpages matching the search request from the plurality of webpage results according to the core information of each webpage, performs cluster analysis on the webpages matching the search request, and generates search results to be displayed to the user. Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • a fourth aspect of the present invention provides a non-volatile computer storage medium storing one or more programs, when the one or more programs are executed by one device, causing the device Performing the following steps: acquiring a plurality of webpages according to a user's search request, performing a specific name recognition word segmentation process on each webpage's title, and acquiring part of speech and extended information corresponding to the word segmentation result;
  • the webpage matching the search request is subjected to cluster analysis, and the search result is generated and displayed to the user.
  • the computer storage medium of the embodiment of the present invention acquires a plurality of web pages according to a search request of the user, performs a special name recognition word segmentation process on the title of each web page, acquires part of speech and extended information corresponding to the word segmentation result, and uses the part of speech corresponding to the word segmentation result. And expanding the information to determine the core information of each webpage, filtering the webpage matching the search request from the plurality of webpage results according to the core information of each webpage, performing cluster analysis on the webpage matching the search request, and generating the search result and displaying to the user . Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • FIG. 1 is a schematic diagram of a conventional search result display interface according to an embodiment of the present invention.
  • FIG. 2 is a flow chart of a method for displaying search results according to an embodiment of the present invention
  • FIG. 3 is a diagram showing the part of speech and the word segmentation corresponding to the word segmentation result by the deep lexical analysis technique according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a search result display interface according to an embodiment of the present invention.
  • FIG. 5 is a flowchart of a method for displaying search results according to an embodiment of the present invention.
  • 6(a)-6(c) are diagrams showing an example of determining core information of each web page based on part of speech and extension information corresponding to the word segmentation result, according to an embodiment of the present invention
  • FIG. 7 is a schematic structural diagram of a search result display apparatus according to an embodiment of the present invention.
  • FIG. 8 is a schematic structural diagram of a search result display apparatus according to an embodiment of the present invention.
  • FIG. 9 is a schematic structural diagram of a search result display apparatus according to another embodiment of the present invention.
  • FIG. 2 is a flow chart of a method of displaying search results in accordance with one embodiment of the present invention.
  • the search result display method includes:
  • S101 Acquire a plurality of webpages according to a search request of the user, perform a special name recognition word segmentation process on the title of each webpage, and obtain part of speech and extended information corresponding to the word segmentation result.
  • the traditional search engine obtains a webpage matching the search request based on the analysis of the words in the webpage content, and provides the plurality of webpages to the user. For example, if the user search request is “Lingdu F8”, the search engine may provide a certain number of webpages whose webpage title contains “Lingdu F8” as a search result on the display interface (such as the screen of the mobile phone). Therefore, it may result in a high repetition rate of the core view of the webpage in the search result provided to the user at a limited real interface, and limited effective information provided to the user.
  • the search result display method of the embodiment of the present invention performs a pragmatic level analysis on a plurality of web pages obtained according to the user's search request. Analyze and understand the basic functions of related vocabulary in the text, such as nouns or verbs, and the fields of related vocabulary, etc., to analyze multiple web pages obtained according to the user's search request, so as to generate more core ideas based on the analysis results. More diversified search results are displayed to users.
  • a plurality of web pages are acquired according to a search request of the user, and a title identification and word segmentation processing are performed on the title of each web page, and the part of speech and extended information corresponding to the word segmentation result are obtained, wherein the part of speech may include a noun, Verbs, adverbs, or proper nouns, etc., extended information may include: name category, name chain analysis, vocabulary field Analysis and so on.
  • proper name category focuses on the category attributes (static attributes) existing in the entity itself, such as people, places, institutions, creatures, articles, virtual works, etc., and the goal is to be able to dynamically identify a certain context in the current context.
  • category attributes static attributes
  • the terminology chain refers to the specific focus of the hot entity. The goal is to find the correct and unambiguous description of the current proper noun in the knowledge base (encyclopedia, freebase, wikipedia, etc.). Nouns are associated by a unique identifier in a knowledge base.
  • the vocabulary domain analysis focuses on application requirements, and its goal is to analyze which vocabulary is more likely to be used by the application context or the website in the current context.
  • the title of each web page is subjected to a special name recognition word segmentation process by an in-depth lexical analysis technique, and the part of speech and extended information corresponding to the word segmentation result are obtained.
  • a deep lexical analysis technique is used to perform a special name recognition word segmentation process for a webpage titled "Spicy Hairy Crabs Over Water", and the part of speech and extended information corresponding to the word segmentation result are obtained as an example, and the detailed lexical analysis technique is described in detail.
  • the deep lexical analysis technique is divided into a shallow analysis layer and a depth analysis layer, and in the shallow analysis layer, focusing on shallow lexical information such as vocabulary classification granularity and lexical function, through the shallow analysis layer.
  • shallow lexical information such as vocabulary classification granularity and lexical function
  • the title of the page "Spicy Hairy Crabs Over Water” is divided into “fragrance”, “spicy”, “hairy crab”, “over”, “water”, and the steps of merging and granularity can be completed in one step through integration and appropriate adjustment. Identifying tasks such as “fragrance”, “spicy”, “hairy crab”, “over” and “water” are combined into the larger-sized segmentation “Spicy Hairy Crab”, “Over Water”, and “Spicy Hairy Crab”. The word “over water”.
  • the special category (such as the traditional PER, LOC, ORG and the current search engine-specific food, video), etc. are abstracted as a generic class Z.
  • the special category such as the traditional PER, LOC, ORG and the current search engine-specific food, video
  • the depth analysis layer deeply analyzes the results of the large-grained word segmentation obtained by the shallow analysis layer, and the result of the word segmentation of the "sweet hairy crab” and “water”, and obtains the special name category and the corresponding corresponding to the word segmentation result.
  • the name chain refers to extended information such as analysis and vocabulary domain analysis. As shown in Figure 3, the extended information of the "sweet hairy crab” obtained is "food”, “biological”, “book”, “culture”, “food” and the like.
  • each webpage title is subjected to a proper name recognition analysis process, and then the part of speech and extended information corresponding to each proper name are obtained, and then each specialization is obtained.
  • Name and its corresponding part of speech and extension information are "sticked" to obtain the proper name of each page title and its corresponding part of speech and extension letter interest.
  • the second example completes the process of identifying the word segmentation of the title of the webpage and obtaining the part of speech and extended information corresponding to the segmentation result in one time, that is, obtaining the corresponding word segmentation processing of the title of the webpage, and acquiring the corresponding word segmentation The part of speech and extended information.
  • the title of the webpage is firstly recognized and the word segmentation corresponding to the word segmentation result is obtained, and then the extension information corresponding to the word segmentation is obtained according to the processing result. .
  • the word segmentation and extension information corresponding to the obtained word segmentation result include information such as a noun, a verb, and a vocabulary field
  • the vocabulary related to the page title is not included in the page title for the various vocabulary words in the page title.
  • the vocabulary of the actual word attribute, such as the verb "" is filtered out, or the vocabulary with the part of speech and extended information is filtered and retained, and then the vocabulary in the title of the web page with the actual part of speech attribute, the part of speech and the extended information are not duplicated.
  • the corresponding part of speech and extended information are determined as the core information of each web page.
  • the "driving recorder” and its related information can be filtered, and the core information of the webpage is determined to be the Lingdu F8 and its corresponding part of speech and extended information.
  • S104 Perform cluster analysis on the webpage matching the search request, and generate a search result to display to the user.
  • the user is provided with richer information as much as possible, and the search request is provided on the display interface.
  • the matched webpages are clustered and analyzed, and the webpages with the same core views are grouped into one class, and the repeated webpages of the core viewpoints are repeatedly displayed to the user, and then the search results are generated and displayed to the user according to the cluster analysis result, in the search results, Multiple web pages of the same cluster display only one of the webpage related information.
  • the search result includes webpage information of a plurality of categories of core viewpoints. As shown in FIG. 4, the webpage matching the search request “Lingdu F8” is clustered and analyzed, and the search result displayed to the user is about the Lingdu F8. Search results for various categories of core views, such as price, evaluation, installation, and questions.
  • the display interface includes more rich webpage information at a time, and when only one webpage related information is displayed for multiple webpages of the same cluster, other A plurality of webpages belonging to the cluster are hidden, and the hidden event is connected with the related triggering menu, so that the user can obtain related information of other hidden webpages of the same cluster by operating the related triggering menu. For example, the user can obtain other hidden webpage information of the cluster by clicking the “+” menu in the upper right corner of a clustering webpage area in the search result.
  • the search result display method of the embodiment of the present invention acquires a plurality of web pages according to a search request of the user, performs a special name recognition word segmentation process on the title of each web page, and obtains part of speech and extended information corresponding to the word segmentation result, according to The part of speech and the extended information corresponding to the word segmentation result determine the core information of each webpage, and the webpage matching the search request is filtered from the plurality of webpage results according to the core information of each webpage, and the webpage matching the search request is clustered and analyzed. Generate search results to show to users. Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • the core information is used as the core proper noun and the core verb in conjunction with FIG. 5 and FIG. 6(a)-(c).
  • the core concept words are taken as an example to illustrate a method for displaying search results according to an embodiment of the present invention, which is described as follows:
  • FIG. 5 is a flowchart of a method for displaying search results according to an embodiment of the present invention. As shown in FIG. 5, the method includes:
  • S501 Acquire a plurality of webpages according to a search request of the user, perform a special name recognition word segmentation process on the title of each webpage, and obtain part of speech and extended information corresponding to the word segmentation result.
  • the core proper noun is the name of the entity mainly described by the web page title
  • the core verb is the main action of the entity mainly described in the web page title.
  • Apple for a webpage titled “Apple's Method of Planting", its core term is “Apple” and its core verb is "planting.”
  • the core proper nouns and core verbs of each web page are determined according to the part of speech and extended information corresponding to the word segmentation result, and the core viewpoints of the web page description are extracted.
  • FIG. 6(a)-FIG.6(c) the core information of extracting the core information for a webpage is illustrated in conjunction with FIG. 6(a)-FIG.6(c).
  • Workflow As shown in Figure 6(a), based on the deep lexical analysis technique, the title of the webpage is lexically labeled. The title of the webpage "Ling F8 driving recorder does not start with the car one night - Baidu knows” The participle and extension information of the word segmentation processing and labeling each participle is: “Lingdu F8" is the core proper noun.
  • the core view of the web page is as shown in FIG. 6(c), as shown in FIG. 6(c), the web page and the core proper nouns. It is "Lingdu F8", and the core verb is "Do not start with the car”.
  • the title of the webpage sometimes does not match the content of the webpage, for example, the title of the webpage is "Apple.
  • the webpage content of the "planting method” may mainly describe "the origin of apple", so it is necessary to detect whether the corresponding webpage content contains core proper nouns and core verbs according to the core information of each webpage to judge the content of the webpage. Whether to describe the information related to the title of the page.
  • the webpage contains the core proper nouns and the core verbs
  • TF-IDF Term Frequency–inverse Document Frequency
  • TF-IDF is a statistical method used to evaluate a word for a file set or one of the files in a corpus.
  • the importance of TF-IDF can weaken the influence of common words, such as the influence of commonly used auxiliary words such as "", and accurately assess the importance of core verbs in the content of web pages.
  • the TF-IDF of the core verb is higher than a preset threshold, it indicates that the content related to the webpage title is described in detail in the content of the webpage, such as a webpage content titled “Apple's Planting Method”, detailed Describes the apple cultivation method.
  • the web page with the core verb's TF-IDF higher than the preset threshold is some high quality web pages.
  • the system in order to maximize the possibility of filtering out the hitting of the noise webpage and ensuring the recall of the related webpage, the system can recall more diversity under the premise of the same number of recalls.
  • the result of the page detecting whether the core proper noun is consistent with the search request, and if the core proper noun is consistent with the search request, determining that the web page is a web page that matches the search request.
  • the clustering metric function is used to analyze the participating clustering.
  • the combined similarity between the core concept words and the core verbs in the webpage If the combined similarity is greater than the preset threshold, it indicates that the webpage participating in the cluster describes the same aspect of the entity corresponding to the core proper noun, thereby participating in the clustering. Web page Perform clustering.
  • the core concept words and core verbs in the web page can represent which aspects of the entities described in the content of the web page.
  • the core concept word can be a noun concept consistent with the core proper noun.
  • the core concept word is “driving recorder”, which is consistent with the core proper noun “Lingdu F8”. Both describe the "Lingdu F8", so that the core concept words are the same as the core proper nouns, and the clusters can be deleted when calculating the similarity between the core concept words and the core verbs in the above web pages. Concept words, only the similarity of the core verbs is calculated.
  • the core concept words can also be noun concepts that are inconsistent with the core proper nouns.
  • the title of the page is “BMW X5 engine ignition”, and its core concept word is “engine”, which is inconsistent with the core proper term “BMW X5”.
  • the core concept words need to be strictly preserved, and the core concept words and core verbs are calculated. The combined similarity.
  • the core concept words and the core verbs are displayed to the user as search result titles, so that the user can more clearly understand the core information of the search results.
  • the search result display method of the embodiment of the present invention detects whether the corresponding webpage content includes a core proper noun and a core verb according to the core information of each webpage, and the size and user search of the TF-IDF of the core verb. Request matching web pages, and display and display the web pages with the same core similarity and core concept words and core verbs with higher similarity, further ensuring that the core views are more abundant and diversified. Search results.
  • FIG. 7 is a schematic structural diagram of a search result display device according to an embodiment of the present invention.
  • the search result display apparatus includes: a first acquisition module 100, a second acquisition module 200, a screening module 300, and a cluster generation module 400.
  • the first obtaining module 100 is configured to acquire a plurality of web pages according to a search request of the user, perform a special name recognition word segmentation process on the title of each web page, and obtain part of speech and extended information corresponding to the word segmentation result.
  • the first obtaining module 100 acquires a plurality of webpages according to a search request of the user, performs a proper name recognition and word segmentation processing on the title of each webpage, and obtains part of speech and extended information corresponding to the word segmentation result, where Part of speech can include nouns, verbs, adverbs, or proper nouns.
  • Extended information can include: proper name categories, proper name chain analysis, and vocabulary domain analysis.
  • proper name category focuses on the category attributes (static attributes) existing in the entity itself, such as people, places, institutions, creatures, articles, virtual works, etc., and the goal is to be able to dynamically identify a certain context in the current context.
  • category attributes static attributes
  • the full name chain refers to the analysis of the specific focus of the hot entity, the goal is in the knowledge base (encyclopedia, freebase, Wikipedia, etc.) finds the correct, unambiguous description of the current proper nouns, and associates related nouns with a unique identifier in a knowledge base.
  • the vocabulary domain analysis focuses on application requirements, and its goal is to analyze which vocabulary is more likely to be used by the application/category website in the current context.
  • the second obtaining module 200 is configured to determine core information of each webpage according to the part of speech and the extended information corresponding to the word segmentation result.
  • the second acquisition module 200 is for a variety of vocabulary in the page title.
  • the vocabulary without the actual word attribute in the web page title may be filtered out, or the vocabulary with the part of speech and the extended information may be filtered and retained, and the vocabulary with the actual part-of-speech attribute and the part of the vocabulary and the extended information may not be duplicated.
  • the vocabulary in the title and its corresponding part of speech and extended information are determined as the core information of each web page.
  • the screening module 300 is configured to filter webpages matching the search request from the plurality of webpage results according to the core information of each webpage.
  • the cluster generation module 400 is configured to perform cluster analysis on the webpage matching the search request, and generate a search result to display to the user.
  • the screening module 300 filters the webpages that match the search request from the plurality of webpage results according to the core information of each webpage, in order to save the display space, the user is provided with richer information as much as possible on the display interface.
  • the class generation module 400 performs cluster analysis on the webpage matching the search request, and aggregates the webpages with the core viewpoints into one category, avoids repeated display of the webpage with repeated core viewpoints to the user, and then generates search results according to the cluster analysis result and presents them to the user. In the search result, only one webpage related information is displayed for a plurality of webpages of the same cluster.
  • the search result display device of the embodiment of the present invention acquires a plurality of web pages according to a search request of the user, performs a special name recognition word segmentation process on the title of each web page, and obtains part of speech and extended information corresponding to the word segmentation result, according to The part of speech and the extended information corresponding to the word segmentation result determine the core information of each webpage, and the webpage matching the search request is filtered from the plurality of webpage results according to the core information of each webpage, and the webpage matching the search request is clustered and analyzed. Generate search results to show to users. Therefore, by performing cluster analysis on the webpage matching the search request, the user is provided with a richer and more diverse search result with a core view.
  • the core information is taken as the core proper noun, the core verb and the core conceptual word as an example, with reference to FIG. 8 and FIG.
  • a search result display device is illustrated as follows:
  • FIG. 8 is a schematic structural diagram of a search result display apparatus according to an embodiment of the present invention.
  • the screening module 300 includes: a first detecting unit 310 and a second detecting unit. 320, a third detecting unit 330 and a determining unit 340.
  • the first detecting unit 310 is configured to detect, according to core information of each webpage, whether the corresponding webpage content includes a core proper noun and a core verb.
  • the webpage content may mainly describe “the origin of the apple”, and thus the first detecting unit 310 It is necessary to detect whether the core properness and the core verb are included in the corresponding webpage content according to the core information of each webpage, so as to determine whether the information related to the webpage title is described in the webpage content.
  • the second detecting unit 320 is configured to detect whether the TF-IDF of the core verb is higher than a preset threshold when the content of the webpage includes the core proper noun and the core verb.
  • the second detecting unit 320 is required to detect the core. Whether the TF-IDF of the verb is higher than a preset threshold, that is, whether there is relatively more content description in the content of the webpage related to the content of the webpage title.
  • the third detecting unit 330 is configured to detect whether the core proper noun is consistent with the search request when the TF-IDF of the core verb is higher than a preset threshold.
  • the determining unit 340 is configured to determine, when the core proper noun is consistent with the search request, a webpage that matches the search request.
  • the TF-IDF of the core verb is higher than a preset threshold, it indicates that the content related to the webpage title is described in detail in the content of the webpage, such as a webpage content titled “Apple's Planting Method”, detailed Describes the apple cultivation method.
  • the web page with the core verb's TF-IDF higher than the preset threshold is some high quality web pages.
  • the system can recall more diversity under the premise of the same number of recalls.
  • the third detecting unit 330 detects whether the core proper noun is consistent with the search request, and if the core proper noun is consistent with the search request, the determining unit 340 determines that the web page is a web page that matches the search request.
  • the cluster generation module 400 includes: a fourth detection unit 410, The analyzing unit 420, the clustering unit 430, and the display unit 440.
  • the fourth detecting unit 410 is configured to detect whether the core proper nouns in the participating webpages are consistent.
  • the fourth detecting unit 410 detects whether the core proper nouns in the participating clustering webpages are consistent. .
  • the analyzing unit 420 is configured to analyze the combined similarity of the core concept words and the core verbs in the clustering webpage by using the clustering metric function when the core proper nouns in the clustering webpage are consistent.
  • the clustering unit 430 is configured to cluster the participating cluster pages when the combined similarity is greater than a preset threshold.
  • the analyzing unit 420 analyzes the clustering metric function. Participating in the combined similarity between the core concept words and the core verbs in the clustering webpage. If the combined similarity is greater than the preset threshold, it indicates that the webpage participating in the clustering describes the same aspect of the entity corresponding to the core proper noun, thereby Class unit 430 will participate in clustering web pages for clustering.
  • the display unit 440 is configured to display the core concept words and the core verbs as search result titles to the user.
  • the display unit 440 displays the core concept words and the core verbs as search result titles to the user, so that the user can more clearly understand the core information of the search results.
  • search result display device described in the embodiment of the present invention corresponds to the search result display method described above with reference to FIG. 1 to FIG. 6 , and details that are not disclosed in the embodiment of the present invention are not described herein again.
  • the search result display apparatus detects, according to the core information of each webpage, whether the corresponding webpage content includes a core proper noun and a core verb, and the size of the TF-IDF of the core verb is obtained and the user searches.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Information Transfer Between Computers (AREA)
  • Machine Translation (AREA)

Abstract

一种搜索结果展示方法和装置，其中方法包括：根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息（S101）；根据与分词结果对应的词性和扩展信息确定每个网页的核心信息（S102）；根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页（S103）；将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户（S104）。该方法对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。

Description

搜索结果展示方法和装置
相关申请的交叉引用
本发明要求北京百度网讯科技有限公司于2016年6月30日提交的、发明名称为“搜索结果展示方法和装置”的、中国专利申请号“201610509333.6”的优先权。
技术领域
本发明涉及信息处理技术领域，尤其涉及一种搜索结果展示方法和装置。
背景技术
随着互联网技术的发展，用户将搜索引擎看为是检索网络信息的主要途径，比如，用户可通过的在搜索引擎中搜索“凌度F8”，以获取与“凌度F8”相关的搜索结果。
然而，相关技术中，比如如图1所示，用于在搜索引擎中输入“凌度F8”后，搜索引擎根据“凌度F8”获取的搜索结果集中于关于“凌度F8”的评测上，因而搜索引擎仅仅为用户提供“凌度F8”的评测和价格等信息，搜索引擎提供的搜索结果并不够多样化且重复率高。
发明内容
本发明的目的旨在至少在一定程度上解决上述的技术问题之一。
为此，本发明的第一个目的在于提出一种搜索结果展示方法，该方法对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
本发明的第二个目的在于提出一种搜索结果展示方法装置。
本发明的第三个目的在于提出一种设备。
本发明的第四个目的在于提出一种非易失性计算机存储介质。
为了实现上述目的，本发明第一方面实施例提出了一种搜索结果展示方法，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
本发明实施例的搜索结果展示方法，根据用户的搜索请求获取多个网页，对每个网页 的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
另外，本发明实施例的搜索结果展示方法，还具有如下附加的技术特征：
在本发明的一个实施例中，所述词性包括：名词、动词、副词、或，专有名词；
所述扩展信息包括：专名类别、专名链指分析、或，词汇领域分析。
在本发明的一个实施例中，所述核心信息包括：核心专有名词和核心动词；
所述根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页，包括：
根据每个网页的核心信息检测对应的网页内容是否包含所述核心专有名词和所述核心动词；
如果所述网页内容包含所述核心专有名词和所述核心动词，检测所述核心动词的TF-IDF是否高于预设阈值；
如果所述核心动词的TF-IDF高于预设阈值，检测所述核心专有名词与所述搜索请求是否一致；
如果所述核心专有名词与所述搜索请求一致，确定是与所述搜索请求匹配的网页。
在本发明的一个实施例中，所述核心信息还包括：核心概念词；
所述将与所述搜索请求匹配的网页进行聚类分析，包括：
检测参与聚类网页中的核心专有名词是否一致；
如果参与聚类网页中的核心专有名词一致，通过聚类度量函数分析参与聚类网页中的所述核心概念词和所述核心动词的组合相似度；
如果所述组合相似度大于预设阈值，则将参与聚类网页进行聚类。
在本发明的一个实施例中，所述生成搜索结果展示给用户，包括：
将所述核心概念词和所述核心动词作为所述搜索结果标题展示给用户。
本发明第二方面的实施例提出了一种搜索结果展示装置，包括：
第一获取模块，用于根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
第二获取模块，用于根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
筛选模块，用于根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
聚类生成模块，用于将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
本发明实施例的搜索结果展示装置，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
另外，本发明实施例的搜索结果展示装置还具有如下附加的技术特征：
在本发明的一个实施例中，所述词性包括：名词、动词、副词、或，专有名词；
所述扩展信息包括：专名类别、专名链指分析、或，词汇领域分析。
在本发明的一个实施例中，所述核心信息包括：核心专有名词和核心动词；
所述筛选模块包括：
第一检测单元，用于根据每个网页的核心信息检测对应的网页内容是否包含所述核心专有名词和所述核心动词；
第二检测单元，用于在所述网页内容包含所述核心专有名词和所述核心动词时，检测所述核心动词的TF-IDF是否高于预设阈值；
第三检测单元，用于在所述核心动词的TF-IDF高于预设阈值时，检测所述核心专有名词与所述搜索请求是否一致；
确定单元，用于在所述核心专有名词与所述搜索请求一致时，确定是与所述搜索请求匹配的网页。
在本发明的一个实施例中，所述核心信息还包括：核心概念词；
所述聚类生成模块包括：
第四检测单元，用于检测参与聚类网页中的核心专有名词是否一致；
分析单元，用于在参与聚类网页中的核心专有名词一致时，通过聚类度量函数分析参与聚类网页中的所述核心概念词和所述核心动词的组合相似度；
聚类单元，用于在所述组合相似度大于预设阈值时，将参与聚类网页进行聚类。
在本发明的一个实施例中，所述聚类生成模块还包括：
展示单元，用于将所述核心概念词和所述核心动词作为所述搜索结果标题展示给用户。
本发明第三方面实施例提供了一种设备，包括：一个或者多个处理器；存储器；一个或者多个程序，所述一个或者多个程序存储在所述存储器中，当被所述一个或者多个处理器执行时，执行以下步骤：根据用户的搜索请求获取多个网页，对每个网页的标 题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
本发明实施例的设备，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
本发明第四方面实施例提供了一种非易失性计算机存储介质，所述计算机存储介质存储有一个或者多个程序，当所述一个或者多个程序被一个设备执行时，使得所述设备执行以下步骤：根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
本发明实施例的计算机存储介质，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
本发明附加的方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本发明的实践了解到。
附图说明
本发明上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1是根据本发明一个实施例的传统搜索结果展示界面示意图；
图2是根据本发明一个实施例的搜索结果展示方法的流程图；
图3是根据本发明一个实施例的通过深度词法分析技术获取与分词结果对应的词性和 扩展信息的示例图；
图4是根据本发明一个实施例的搜索结果展示界面示意图
图5是根据本发明一个具体实施例的搜索结果展示方法的流程图；
图6(a)-图6(c)是根据本发明一个实施例的根据与分词结果对应的词性和扩展信息确定每个网页的核心信息的示例图；
图7是根据本发明一个实施例的搜索结果展示装置的结构示意图；
图8是根据本发明一个具体实施例的搜索结果展示装置的结构示意图；以及
图9是根据本发明另一个实施例的搜索结果展示装置的结构示意图。
具体实施方式
下面详细描述本发明的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，旨在用于解释本发明，而不能理解为对本发明的限制。
下面参考附图描述本发明实施例的搜索结果的展示方法和装置。
图2是根据本发明一个实施例的搜索结果展示方法的流程图。
如图2所示，该搜索结果展示方法包括：
S101，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息。
通常，传统的搜索引擎在根据用户的搜索请求获取多个网页后，仅仅基于对网页内容中的用词的分析，获取与搜索请求匹配的网页，并将该多个网页提供给用户。比如，用户搜索请求为“凌度F8”,搜索引擎可能会在展示界面上(比如手机的屏幕上)将一定数目的、网页标题包含“凌度F8”的网页作为搜索结果提供给用户。从而，可能会导致在有限的真是界面上，一次提供给用户的搜索结果中的网页核心观点重复率高，为用户提供的有效信息有限。
为了使得在有限的展示界面上，一次提供给用户的搜索结果更加丰富多样化，本发明实施例的搜索结果展示方法，对根据用户的搜索请求获取的多个网页进行基于语用层面的分析，通过分析理解相关词汇在文本中扮演的基本功能，比如是名词还是动词等，以及相关词汇的领域等来对根据用户的搜索请求获取的多个网页进行分析，从而根据分析结果生成核心观点更加丰富、更加多元化的搜索结果展示给用户。
具体地，在实际应用中，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别和分词处理，获取与分词结果对应的词性和扩展信息，其中，词性可以包括名词、动词、副词、或，专有名词等，扩展信息可包括：专名类别、专名链指分析、词汇领域分 析等。
应当理解的是，上述专名类别重点关注实体本身所存在的类别属性(静态属性)，如人、地、机构、生物、物品、虚拟作品等，目标是能够动态识别出在当前语境中某一专有名词的具体类别。
专名链指分析重点关注热门实体的具体所指，其目标是在知识库(百科、freebase、wikipedia等)中寻找到当前专有名词的正确、无歧义描述页面(entry)，将相关专有名词通过某一知识库唯一标识关联起来。
词汇领域分析重点关注应用需求，其目标是分析出任意词汇在当前语境下，更倾向于被哪一种应用场景或垂类网站所使用。
基于以上描述，可以理解，本发明实施例的搜索结果展示方法中，通过一种深度词法分析技术对每个网页的标题进行专名识别分词处理，并获取与分词结果对应的词性和扩展信息，下面结合附图3以通过深度词法分析技术对一个网页标题为“香辣大闸蟹过水”进行专名识别分词处理，获取与分词结果对应的词性和扩展信息为例，详细说明深度词法分析技术的工作流程：
如图3所示，该深度词法分析技术分为浅层分析层和深度分析层，在浅层分析层，重点关注词汇的分类粒度、词法功能等浅层的词法信息，经过该浅层分析层将网页标题“香辣大闸蟹过水”分词为“香”、“辣”、“大闸蟹”、“过”、“水”，并可通过整合与适当调整等步骤，一次性完成粒度合并与的词性识别等任务，即将“香”、“辣”、“大闸蟹”、“过”、“水”合并为粒度较大的分词“香辣大闸蟹”、“过水”，并获取“香辣大闸蟹”、“过水”的词性。
其中，如图3所示，在浅层分析层的处理过程中，专名类别(如传统的PER、LOC、ORG以及当前搜索引擎特有的美食、视频)等被整体抽象为专名大类Z，以便减轻目标空间的负荷，并将目标聚焦至浅层词法分析部分。
进而，深度分析层将浅层分析层得到的大粒度的分词结果，即将得到的“香辣大闸蟹”、“过水”的分词结果进行深度分析，获取与该分词结果对应的专名类别、专名链指分析、词汇领域分析等扩展信息，如图3所示，获取的“香辣大闸蟹”的扩展信息为“食物”、“生物”“书籍”、“养殖”、“美食”等。
需要说明的是，根据具体应用场景的不同，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息的方式不同：
第一种示例，在根据用户的搜索请求获取多个网页后，先对每个网页标题进行专名识别分析处理，之后再获取与每个专名对应的词性和扩展信息，进而将每个专名和与其对应的词性和扩展信息进行“黏贴”，以获取每个网页标题的专名以及与其对应的词性和扩展信 息。
第二种示例，一次性完成对网页的标题进行专名识别分词处理和获取与分词结果对应的词性和扩展信息，即在完成对网页的标题进行专名识别分词处理的同时，获取与分词对应的词性和扩展信息。
第三种示例，如图3所示的处理方法中，先一次性完成对网页的标题进行专名识别分词处理和获取与分词结果对应的词性，进而根据处理结果再获取分词对应的扩展信息等。
S102，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息。
具体地，由于获取的分词结果对应的词性和扩展信息中，包括网页标题相关词汇的是名词、动词以及属于哪个词汇领域等信息，针对网页标题中多种多样的词汇，可将网页标题中没有实际词属性的词汇，如助词“的”过滤掉，或者将词性和扩展信息重复的词汇过滤并保留一个等，进而将具有实际词性属性的、词性和扩展信息不重复的网页标题中的词汇及其对应的词性和扩展信息确定为每个网页的核心信息。
比如，对于网页标题为“凌度F8行车记录仪”的网页进行分析处理后，获取分词处理后获取两个分词“凌度F8”和“行车记录仪”，而由于凌度F8即为行车记录仪，因而，可将“行车记录仪”及其相关信息过滤，并确定该网页的核心信息为凌度F8及其对应的词性和扩展信息。
S103，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页。
S104，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
具体地，根据每个网页的核心信息从多个网页结果中筛选出与搜索请求匹配的网页后，为了节约展示空间，在展示界面上尽可能的为用户提供更丰富的信息，将与搜索请求匹配的网页进行聚类分析，将核心观点一致的网页聚成一类，避免核心观点重复的网页重复展示给为用户，进而根据聚类分析结果生成搜索结果展示给用户，在该搜索结果中，针对同一聚类的多个网页只显示其中一个网页相关信息。
其中，该搜索结果包含多种类别的核心观点的网页信息，如图4所示，将与搜索请求“凌度F8”匹配的网页进行聚类分析，展示给用户的搜索结果是关于凌度F8的价格、评测、安装、问题等多种类别的核心观点的搜索结果。
在本发明的一个实施例中，为了进一步的丰富搜索结果，使得展示界面上一次包含更丰富的网页信息，在针对同一聚类的多个网页只显示其中一个网页相关信息时，还可将其他属于该聚类的多个网页进行隐藏，并将该隐藏事件与相关触发菜单进行连接，从而用户可通过对相关触发菜单的操作，获取同一聚类的其他隐藏网页的相关信息。比如，用户可通过点击搜索结果中的展示某个聚类网页区域右上角的“+”菜单，获取该聚类的其他隐藏的网页信息。
综上所述，本发明实施例的搜索结果展示方法，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
基于以上实施例，进一步地，为了更加清楚的说明本发明实施例的搜索结果展示方法，下面结合附图5和图6(a)-(c)，以核心信息为核心专有名词、核心动词和核心概念词为例，举例说明本发明实施例的搜索结果展示方法，说明如下：
图5是根据本发明一个具体实施例的搜索结果展示方法的流程图，如图5所示，该方法包括：
S501，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息。
S502，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，其中，核心信息包括：核心专有名词和核心动词。
可以理解，核心专有名词是网页标题主要描述的实体的名称等，核心动词是该网页标题中主要描述的实体的主要动作等。比如，对于一篇标题为“苹果的种植方法”的网页，其核心专有名词是“苹果”，其核心动词是“种植”。
具体地，根据与分词结果对应的词性和扩展信息确定每个网页的核心专有名词和核心动词，抽取到该网页描述的核心观点。
为了更加清楚的描述本发明实施例的如何根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，下面结合图6(a)-图6(c)举例说明对一个网页提取核心信息的工作流程：如图6(a)所示，基于深度词法分析技术，对该网页的标题进行词法标注，对该网页的标题“凌F8行车记录仪过一晚上不随车启动-百度知道”，进行分词处理并标注各个分词的词性和扩展信息为：“凌度F8”为核心专有名词等。
进而，从网页标题的各个分词的词性和扩展信息中抽取出具有实际词性属性或者专有名词类型的词，得到如图6(b)所示的结果。
进一步的，在抽取出具有实际词性属性或者专有名词类型的词后，可以确定网页的核心观点如图6(c)所示，如图6(c)所示，该网页和核心专有名词是“凌度F8”,核心动词为“不随车启动”等。
S503，检测对应的网页内容是否包含核心专有名词和核心动词。
具体地，由于网页的标题与网页的内容有时候并不相符，比如对于网页标题为“苹果 的种植方法”的网页，其网页内容中可能主要描述“苹果的产地”，因此需要根据每个网页的核心信息检测对应的网页内容中是否包含核心专有名词和核心动词，以判断网页内容中是否描述了与网页标题相关的信息。
S504，如果网页内容包含核心专有名词和核心动词，检测核心动词的TF-IDF是否高于预设阈值。
具体地，当判断网页中包含核心专有名词和核心动词后，为了进一步保证网页内容中描述的内容和网页标题的相关性大，保证获取的网页的质量，还需要检测核心动词的TF-IDF是否高于预设阈值，即检测网页内容中是否有相对较多的内容描述与网页标题相关的内容。
其中，TF-IDF(Term Frequency–inverse Document Frequency,信息检索与数据挖掘的常用加权技术)，是一种统计方法，用以评估一字词对于一个文件集或一个语料库中的其中一份文件的重要程度，通过TF-IDF可以削弱常用词的影响，比如“的”等常用助词的影响，准确评估核心动词在网页内容中的重要程度。
S505，如果核心动词的TF-IDF高于预设阈值，检测核心专有名词与搜索请求是否一致。
S506，如果核心专有名词与搜索请求一致，确定是与搜索请求匹配的网页。
具体地，如果核心动词的TF-IDF高于预设阈值，则表明该网页内容中详细描述了与网页标题相关的内容，比如一篇网页标题为“苹果的种植方法”的网页内容中，详细描述了苹果的种植方法，核心动词的TF-IDF高于预设阈值的网页是质量较高的一些网页。
进一步地，在上述质量较高的网页中，为了最大限度的过滤掉命中噪音网页的可能性并保证相关网页的召回，从而以保证系统在召回数量不变的前提下，能够召回更多样性的页面结果，检测核心专有名词是否与搜索请求一致，如果核心专有名词与搜索请求一致，则确定该网页是与搜索请求匹配的网页。
S507，检测参与聚类网页中的核心专有名词是否一致。
具体地，为了对获取的与搜索请求匹配的网页进行准确的聚类分析，避免将核心不一致的网页聚类为一类，检测参与聚类网页中的核心专有名词是否一致。
S508，如果参与聚类网页中的核心专有名词一致，通过聚类度量函数分析参与聚类网页中的核心概念词和核心动词的组合相似度。
S509，如果组合相似度大于预设阈值，则将参与聚类网页进行聚类。
具体地，如果参与聚类网页中的核心专有名词一致，则表明网页描述的是一个实体，从而为了进一步区分网页描述的是核心专有名词的哪些方面，通过聚类度量函数分析参与聚类网页中的核心概念词和核心动词的组合相似度，如果组合相似度大于预设阈值，则表明参与聚类的网页描述的是核心专有名词对应的实体的同一个方面，从而将参与聚类网页 进行聚类。
其中，网页中的核心概念词和核心动词可以表征该网页内容中描述的实体的哪些方面。
核心概念词可以是与核心专有名词一致的名词概念，比如如图6(c)所示，核心概念词为“行车记录仪”，其与核心专有名词“凌度F8”一致，二者均是描述了“凌度F8”，从而，对于核心概念词是与核心专有名词一致的参与聚类的网页，在计算上述网页中的核心概念词和核心动词的相似度时，可以删除核心概念词，只计算核心动词的相似度。
核心概念词也可以是与核心专有名词不一致的名词概念，比如对网页标题为“宝马X5发动机点火”，其核心概念词为“发动机”，其与核心专有名词“宝马X5”不一致，从而，对于核心概念词是与核心专有名词不一致的参与聚类的网页，在计算上述网页中的核心概念词和核心动词的相似度时，需要严格保留核心概念词，计算核心概念词和核心动词的组合相似度。
S510，将核心概念词和核心动词作为搜索结果标题展示给用户。
具体地，在本发明的一个实施例中，将核心概念词与核心动词作为搜索结果标题展示给用户，以供用户更加清楚的了解该搜索结果的核心信息。
综上所述，本发明实施例的搜索结果展示方法，根据每个网页的核心信息检测对应的网页内容是否包含核心专有名词和核心动词，以及核心动词的TF-IDF的大小获取与用户搜索请求匹配的网页，并将核心专有名词一致的、且核心概念词和核心动词的组合相似度较高的网页进行具备并展示给用户，进一步保证了为用户提供核心观点更丰富、更加多样化的搜索结果。
为了实现上述实施例，本发明还提出了一种搜索结果展示装置，图7是根据本发明一个实施例的搜索结果展示装置的结构示意图。如图7所示，该搜索结果展示装置包括：第一获取模块100、第二获取模块200、筛选模块300和聚类生成模块400。
其中，第一获取模块100用于根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息。
具体地，在实际应用中，第一获取模块100根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别和分词处理，获取与分词结果对应的词性和扩展信息，其中，词性可以包括名词、动词、副词、或，专有名词等，扩展信息可包括：专名类别、专名链指分析、词汇领域分析等。
应当理解的是，上述专名类别重点关注实体本身所存在的类别属性(静态属性)，如人、地、机构、生物、物品、虚拟作品等，目标是能够动态识别出在当前语境中某一专有名词的具体类别。
专名链指分析重点关注热门实体的具体所指，其目标是在知识库(百科、freebase、 wikipedia等)中寻找到当前专有名词的正确、无歧义描述页面(entry)，将相关专有名词通过某一知识库唯一标识关联起来。
词汇领域分析重点关注应用需求，其目标是分析出任意词汇在当前语境下，更倾向于被哪一种应用场景/垂类网站所使用。
第二获取模块200，用于根据与分词结果对应的词性和扩展信息确定每个网页的核心信息。
具体地，由于获取的分词结果对应的词性和扩展信息中，标注了网页标题相关词汇的是名词、动词以及属于哪个词汇领域等信息，针对网页标题中多种多样的词汇，第二获取模块200可将网页标题中没有实际词属性的词汇如助词“的”过滤掉，或者将词性和扩展信息重复的词汇过滤并保留一个等，进而将具有实际词性属性的、词性和扩展信息不重复的网页标题中的词汇及其对应的词性和扩展信息确定为每个网页的核心信息。
筛选模块300用于根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页。
聚类生成模块400，用于将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
具体地，筛选模块300根据每个网页的核心信息从多个网页结果中筛选出与搜索请求匹配的网页后，为了节约展示空间，在展示界面上尽可能的为用户提供更丰富的信息，聚类生成模块400将与搜索请求匹配的网页进行聚类分析，将核心观点一致的网页聚成一类，避免核心观点重复的网页重复展示给为用户，进而根据聚类分析结果生成搜索结果展示给用户，在该搜索结果中，针对同一聚类的多个网页只显示其中一个网页相关信息。
综上所述，本发明实施例的搜索结果展示装置，根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息，根据与分词结果对应的词性和扩展信息确定每个网页的核心信息，根据每个网页的核心信息从多个网页结果中筛选与搜索请求匹配的网页，将与搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。由此通过对与搜索请求匹配的网页进行聚类分析，为用户提供核心观点更丰富、更加多样化的搜索结果。
基于以上实施例，进一步地，为了更加清楚的说明本发明实施例的搜索结果展示装置，下面结合附图8和图9，以核心信息为核心专有名词、核心动词和核心概念词为例，举例说明本发明实施例的搜索结果展示装置，说明如下：
图8是根据本发明一个具体实施例的搜索结果展示装置的结构示意图，如图8所示，在如图7所示的基础上，筛选模块300包括：第一检测单元310、第二检测单元320、第三检测单元330和确定单元340。
其中，第一检测单元310，用于根据每个网页的核心信息检测对应的网页内容是否包含核心专有名词和核心动词。
具体地，由于网页的标题与网页的内容有时候并不相符，比如对于网页标题为“苹果的种植方法”的网页，其网页内容中可能主要描述“苹果的产地”，因此第一检测单元310需要根据每个网页的核心信息检测对应的网页内容中是否包含核心专有名词和核心动词，以判断网页内容中是否描述了与网页标题相关的信息。
第二检测单元320，用于在网页内容包含核心专有名词和核心动词时，检测核心动词的TF-IDF是否高于预设阈值。
具体地，当判断网页中包含核心专有名词和核心动词后，为了进一步保证网页内容中描述的内容和网页标题的相关性大，保证获取的网页的质量，还需要第二检测单元320检测核心动词的TF-IDF是否高于预设阈值，即检测网页内容中是否有相对较多的内容描述与网页标题相关的内容。
第三检测单元330，用于在核心动词的TF-IDF高于预设阈值时，检测核心专有名词与搜索请求是否一致。
确定单元340，用于在核心专有名词与搜索请求一致时，确定是与搜索请求匹配的网页。
具体地，如果核心动词的TF-IDF高于预设阈值，则表明该网页内容中详细描述了与网页标题相关的内容，比如一篇网页标题为“苹果的种植方法”的网页内容中，详细描述了苹果的种植方法，核心动词的TF-IDF高于预设阈值的网页是质量较高的一些网页。
进一步地，在上述质量较高的网页中，为了最大限度的过滤掉命中噪音网页的可能性并保证相关网页的召回，从而以保证系统在召回数量不变的前提下，能够召回更多样性的页面结果，第三检测单元330检测核心专有名词是否与搜索请求一致，如果核心专有名词与搜索请求一致，确定单元340则确定该网页是与搜索请求匹配的网页。
进而，图9是根据本发明另一个实施例的搜索结果展示装置的结构示意图，如图9所示，在如图8所示的基础上，聚类生成模块400包括：第四检测单元410、分析单元420、聚类单元430和展示单元440。
其中，第四检测单元410，用于检测参与聚类网页中的核心专有名词是否一致。
具体地，为了对获取的与搜索请求匹配的网页进行准确的聚类分析，避免将核心不一致的网页聚类为一类，第四检测单元410检测参与聚类网页中的核心专有名词是否一致。
分析单元420，用于在参与聚类网页中的核心专有名词一致时，通过聚类度量函数分析参与聚类网页中的核心概念词和核心动词的组合相似度。
聚类单元430，用于在组合相似度大于预设阈值时，将参与聚类网页进行聚类。
具体地，如果参与聚类网页中的核心专有名词一致，则表明网页描述的是一个实体，从而为了进一步区分网页描述的是核心专有名词的哪些方面，分析单元420通过聚类度量函数分析参与聚类网页中的核心概念词和核心动词的组合相似度，如果组合相似度大于预设阈值，则表明参与聚类的网页描述的是核心专有名词对应的实体的同一个方面，从而聚类单元430将参与聚类网页进行聚类。
展示单元440，用于将核心概念词和核心动词作为搜索结果标题展示给用户。
具体地，在本发明的一个实施例中，展示单元440将核心概念词与核心动词作为搜索结果标题展示给用户，以供用户更加清楚的了解该搜索结果的核心信息。
需要说明的是，本发明实施例描述的搜索结果展示装置与以上参照图1-图6描述的搜索结果展示方法对应，本发明实施例中未披露的细节，在此不再赘述。
综上所述，本发明实施例的搜索结果展示装置，根据每个网页的核心信息检测对应的网页内容是否包含核心专有名词和核心动词，以及核心动词的TF-IDF的大小获取与用户搜索请求匹配的网页，并将核心专有名词一致的、且核心概念词和核心动词的组合相似度较高的网页进行具备并展示给用户，进一步保证了为用户提供核心观点更丰富、更加多样化的搜索结果。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不必须针对的是相同的实施例或示例。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
尽管上面已经示出和描述了本发明的实施例，可以理解的是，上述实施例是示例性的，不能理解为对本发明的限制，本领域的普通技术人员在本发明的范围内可以对上述实施例进行变化、修改、替换和变型。

Claims (12)

  1. 一种搜索结果展示方法，其特征在于，包括以下步骤：
    根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
    根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
    根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
    将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
  2. 如权利要求1所述的方法，其特征在于，
    所述词性包括：名词、动词、副词、或，专有名词；
    所述扩展信息包括：专名类别、专名链指分析、或，词汇领域分析。
  3. 如权利要求1或2所述的方法，其特征在于，所述核心信息包括：核心专有名词和核心动词；
    所述根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页，包括：
    检测对应的网页内容是否包含所述核心专有名词和所述核心动词；
    如果所述网页内容包含所述核心专有名词和所述核心动词，检测所述核心动词的TF-IDF是否高于预设阈值；
    如果所述核心动词的TF-IDF高于预设阈值，检测所述核心专有名词与所述搜索请求是否一致；
    如果所述核心专有名词与所述搜索请求一致，确定是与所述搜索请求匹配的网页。
  4. 如权利要求1-3任一所述的方法，其特征在于，所述核心信息还包括：核心概念词；
    所述将与所述搜索请求匹配的网页进行聚类分析，包括：
    检测参与聚类网页中的核心专有名词是否一致；
    如果参与聚类网页中的核心专有名词一致，通过聚类度量函数分析参与聚类网页中的所述核心概念词和所述核心动词的组合相似度；
    如果所述组合相似度大于预设阈值，则将参与聚类网页进行聚类。
  5. 如权利要求1-4任一所述的方法，其特征在于，所述生成搜索结果展示给用户，包括：
    将所述核心概念词和所述核心动词作为所述搜索结果标题展示给用户。
  6. 一种搜索结果展示装置，其特征在于，包括：
    第一获取模块，用于根据用户的搜索请求获取多个网页，对每个网页的标题进行专名 识别分词处理，获取与分词结果对应的词性和扩展信息；
    第二获取模块，用于根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
    筛选模块，用于根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
    聚类生成模块，用于将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
  7. 如权利要求6所述的装置，其特征在于，
    所述词性包括：名词、动词、副词、或，专有名词；
    所述扩展信息包括：专名类别、专名链指分析、或，词汇领域分析。
  8. 如权利要求6或7所述的装置，其特征在于，所述核心信息包括：核心专有名词和核心动词；
    所述筛选模块包括：
    第一检测单元，用于根据每个网页的核心信息检测对应的网页内容是否包含所述核心专有名词和所述核心动词；
    第二检测单元，用于在所述网页内容包含所述核心专有名词和所述核心动词时，检测所述核心动词的TF-IDF是否高于预设阈值；
    第三检测单元，用于在所述核心动词的TF-IDF高于预设阈值时，检测所述核心专有名词与所述搜索请求是否一致；
    确定单元，用于在所述核心专有名词与所述搜索请求一致时，确定是与所述搜索请求匹配的网页。
  9. 如权利要求6-8任一所述的装置，其特征在于，所述核心信息还包括：核心概念词；
    所述聚类生成模块包括：
    第四检测单元，用于检测参与聚类网页中的核心专有名词是否一致；
    分析单元，用于在参与聚类网页中的核心专有名词一致时，通过聚类度量函数分析参与聚类网页中的所述核心概念词和所述核心动词的组合相似度；
    聚类单元，用于在所述组合相似度大于预设阈值时，将参与聚类网页进行聚类。
  10. 如权利要求6-9任一所述的装置，其特征在于，所述聚类生成模块还包括：
    展示单元，用于将所述核心概念词和所述核心动词作为所述搜索结果标题展示给用户。
  11. 一种设备，包括：
    一个或者多个处理器；
    存储器；
    一个或者多个程序，所述一个或者多个程序存储在所述存储器中，当被所述一个 或者多个处理器执行时，执行以下步骤：
    根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
    根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
    根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
    将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
  12. 一种非易失性计算机存储介质，所述计算机存储介质存储有一个或者多个程序，当所述一个或者多个程序被一个设备执行时，使得所述设备执行以下步骤：
    根据用户的搜索请求获取多个网页，对每个网页的标题进行专名识别分词处理，获取与分词结果对应的词性和扩展信息；
    根据与分词结果对应的词性和扩展信息确定每个网页的核心信息；
    根据每个网页的核心信息从所述多个网页结果中筛选与所述搜索请求匹配的网页；
    将与所述搜索请求匹配的网页进行聚类分析，生成搜索结果展示给用户。
PCT/CN2016/096480 2016-06-30 2016-08-24 搜索结果展示方法和装置 Ceased WO2018000557A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/313,206 US10977317B2 (en) 2016-06-30 2016-08-24 Search result displaying method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610509333.6 2016-06-30
CN201610509333.6A CN106202285A (zh) 2016-06-30 2016-06-30 搜索结果展示方法和装置

Publications (1)

Publication Number Publication Date
WO2018000557A1 true WO2018000557A1 (zh) 2018-01-04

Family

ID=57464468

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/096480 Ceased WO2018000557A1 (zh) 2016-06-30 2016-08-24 搜索结果展示方法和装置

Country Status (3)

Country Link
US (1) US10977317B2 (zh)
CN (1) CN106202285A (zh)
WO (1) WO2018000557A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109410005A (zh) * 2018-10-30 2019-03-01 北京筑龙信息技术有限责任公司 一种查价方法以及查价装置
CN110705290A (zh) * 2019-09-29 2020-01-17 新华三信息安全技术有限公司 一种网页分类方法及装置
CN112749546A (zh) * 2021-01-13 2021-05-04 叮当快药科技集团有限公司 医疗语义的检索匹配处理方法和装置
CN112925883A (zh) * 2021-02-19 2021-06-08 北京百度网讯科技有限公司 搜索请求处理方法、装置、电子设备及可读存储介质
US20210191995A1 (en) * 2019-12-23 2021-06-24 97th Floor Generating and implementing keyword clusters
US11957670B2 (en) 2018-06-06 2024-04-16 Minoryx Therapeutics S.L. 5-[[4-[2-[5-(1-hydroxyethyl)pyridin-2-yl]ethoxy]phenyl]methyl]-1,3-thiazolidine-2,4-dione and its salts for use in the treatment of mitochondrial diseases

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106777028A (zh) * 2016-12-08 2017-05-31 北京奇虎科技有限公司 一种基于即时通讯软件的信息筛选方法及装置
CN107220300B (zh) * 2017-05-05 2018-07-20 平安科技（深圳）有限公司 信息挖掘方法、电子装置及可读存储介质
CN111309999B (zh) * 2018-12-11 2023-05-16 阿里巴巴集团控股有限公司 一种互动场景内容的生成方法及装置
CN110516127B (zh) * 2019-08-28 2020-08-11 百度在线网络技术（北京）有限公司 多需求搜索请求的划分方法、装置、电子设备及存储介质
CN110941765A (zh) * 2019-12-04 2020-03-31 青梧桐有限责任公司 搜索意图识别方法、信息搜索方法、装置及电子设备
CN111444405A (zh) * 2020-03-20 2020-07-24 北京三快在线科技有限公司 用于搜索中的用户交互方法、装置、移动终端和存储介质
CN112784006B (zh) * 2020-06-05 2024-07-26 珠海金山办公软件有限公司 一种书籍推荐方法、装置、电子设备及可读存储介质
CN113378097B (zh) * 2021-07-08 2024-05-24 北京安天网络安全技术有限公司 一种web页面展示方法、装置、电子设备及存储介质
CN118866249B (zh) * 2024-09-26 2025-05-16 北京积水潭医院贵州医院 一种用于骨科术后康复训练的辅助推荐系统
CN121188267A (zh) * 2025-09-11 2025-12-23 重庆数字资源集团有限公司 一种基于大模型技术的城市治理数据搜索方法及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040220963A1 (en) * 2003-05-01 2004-11-04 Microsoft Corporation Object clustering using inter-layer links
CN101079031A (zh) * 2006-06-15 2007-11-28 腾讯科技（深圳）有限公司 一种网页主题提取系统和方法
US20100185689A1 (en) * 2009-01-20 2010-07-22 Microsoft Corporation Enhancing Keyword Advertising Using Wikipedia Semantics
CN101788988A (zh) * 2009-01-22 2010-07-28 蔡亮华 信息抓取方法
CN102693304A (zh) * 2012-05-22 2012-09-26 北京邮电大学 一种搜索引擎的反馈信息处理方法及搜索引擎

Family Cites Families (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6944612B2 (en) * 2002-11-13 2005-09-13 Xerox Corporation Structured contextual clustering method and system in a federated search engine
US7617176B2 (en) * 2004-07-13 2009-11-10 Microsoft Corporation Query-based snippet clustering for search result grouping
US20070118514A1 (en) * 2005-11-19 2007-05-24 Rangaraju Mariappan Command Engine
US20110238663A1 (en) * 2008-01-10 2011-09-29 Qin Zhang Search method and system using thinking system
US7899818B2 (en) * 2006-03-29 2011-03-01 A9.Com, Inc. Method and system for providing focused search results by excluding categories
CN101114284B (zh) * 2006-07-27 2010-11-24 阿里巴巴集团控股有限公司 一种显示网页内容相关信息的方法及系统
US9318108B2 (en) * 2010-01-18 2016-04-19 Apple Inc. Intelligent automated assistant
US8019760B2 (en) * 2007-07-09 2011-09-13 Vivisimo, Inc. Clustering system and method
US8359191B2 (en) * 2008-08-01 2013-01-22 International Business Machines Corporation Deriving ontology based on linguistics and community tag clouds
US20130151936A1 (en) * 2011-12-12 2013-06-13 Microsoft Corporation Page preview using contextual template metadata and labeling
US20140006012A1 (en) * 2012-07-02 2014-01-02 Microsoft Corporation Learning-Based Processing of Natural Language Questions
US9558275B2 (en) * 2012-12-13 2017-01-31 Microsoft Technology Licensing, Llc Action broker
CN104462105B (zh) * 2013-09-16 2019-01-22 腾讯科技（深圳）有限公司 中文分词方法、装置和服务器
CN104125209B (zh) * 2014-01-03 2015-09-09 腾讯科技（深圳）有限公司 恶意网址提示方法和路由器
US9342561B2 (en) * 2014-01-08 2016-05-17 International Business Machines Corporation Creating and using titles in untitled documents to answer questions
CN103984705B (zh) * 2014-04-25 2018-05-04 北京奇虎科技有限公司 一种搜索结果的展示方法、装置和系统
US10572206B2 (en) * 2015-08-28 2020-02-25 Vinuth Tulasi System and method for minimizing screen space required for displaying auxiliary content
CN105488024B (zh) * 2015-11-20 2017-10-13 广州神马移动信息科技有限公司 网页主题句的抽取方法及装置
CN106096037B (zh) * 2016-06-27 2025-11-07 北京百度网讯科技有限公司 基于人工智能的搜索结果聚合方法、装置以及搜索引擎
US20190286676A1 (en) * 2018-03-16 2019-09-19 Turbine Corporate Holdings, Inc. Contextual content collection, filtering, enrichment, curation and distribution

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040220963A1 (en) * 2003-05-01 2004-11-04 Microsoft Corporation Object clustering using inter-layer links
CN101079031A (zh) * 2006-06-15 2007-11-28 腾讯科技（深圳）有限公司 一种网页主题提取系统和方法
US20100185689A1 (en) * 2009-01-20 2010-07-22 Microsoft Corporation Enhancing Keyword Advertising Using Wikipedia Semantics
CN101788988A (zh) * 2009-01-22 2010-07-28 蔡亮华 信息抓取方法
CN102693304A (zh) * 2012-05-22 2012-09-26 北京邮电大学 一种搜索引擎的反馈信息处理方法及搜索引擎

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11957670B2 (en) 2018-06-06 2024-04-16 Minoryx Therapeutics S.L. 5-[[4-[2-[5-(1-hydroxyethyl)pyridin-2-yl]ethoxy]phenyl]methyl]-1,3-thiazolidine-2,4-dione and its salts for use in the treatment of mitochondrial diseases
CN109410005A (zh) * 2018-10-30 2019-03-01 北京筑龙信息技术有限责任公司 一种查价方法以及查价装置
CN110705290A (zh) * 2019-09-29 2020-01-17 新华三信息安全技术有限公司 一种网页分类方法及装置
CN110705290B (zh) * 2019-09-29 2023-06-23 新华三信息安全技术有限公司 一种网页分类方法及装置
US20210191995A1 (en) * 2019-12-23 2021-06-24 97th Floor Generating and implementing keyword clusters
US11941073B2 (en) * 2019-12-23 2024-03-26 97th Floor Generating and implementing keyword clusters
CN112749546A (zh) * 2021-01-13 2021-05-04 叮当快药科技集团有限公司 医疗语义的检索匹配处理方法和装置
CN112749546B (zh) * 2021-01-13 2023-01-17 叮当快药科技集团有限公司 医疗语义的检索匹配处理方法和装置
CN112925883A (zh) * 2021-02-19 2021-06-08 北京百度网讯科技有限公司 搜索请求处理方法、装置、电子设备及可读存储介质
CN112925883B (zh) * 2021-02-19 2024-01-19 北京百度网讯科技有限公司 搜索请求处理方法、装置、电子设备及可读存储介质

Also Published As

Publication number Publication date
CN106202285A (zh) 2016-12-07
US10977317B2 (en) 2021-04-13
US20190155846A1 (en) 2019-05-23

Similar Documents

Publication Publication Date Title
US10977317B2 (en) Search result displaying method and apparatus
JP6487201B2 (ja) 推奨ページを生成するための方法及び装置
CN101292238B (zh) 语义主题的自动化丰富呈现的方法和系统
US11328159B2 (en) Automatically detecting contents expressing emotions from a video and enriching an image index
CN103873601B (zh) 一种寻址类查询词的挖掘方法及系统
CN106649778B (zh) 基于深度问答的交互方法和装置
JP2017508214A (ja) 検索推奨の提供
CN106601259A (zh) 一种基于声纹搜索的信息推荐方法及装置
WO2014000576A1 (zh) 一种网络搜索方法及网络搜索系统
CN106708886B (zh) 站内搜索词的显示方法及装置
US20180039675A1 (en) Extended search method and apparatus
US9053186B2 (en) Method and apparatus for detecting and explaining bursty stream events in targeted groups
CN110020104A (zh) 新闻处理方法、装置、存储介质及计算机设备
CN109299277A (zh) 舆情分析方法、服务器及计算机可读存储介质
WO2021129122A1 (zh) 书籍查询页面的展示方法、电子设备及计算机存储介质
CN104598617A (zh) 一种搜索结果展示方法及装置
CN108959329A (zh) 一种文本分类方法、装置、介质及设备
WO2017084205A1 (zh) 一种网络用户身份认证方法及系统
CN103617213A (zh) 识别新闻网页属性特征的方法和系统
TW201717067A (zh) 議題顯示系統、議題顯示方法以及電腦可讀取記錄媒體
WO2016107125A1 (zh) 信息搜索方法及装置
US9965766B2 (en) Method to expand seed keywords into a relevant social query
CN108304452A (zh) 文章处理方法及装置、存储介质
JP6042790B2 (ja) トレンド分析装置、トレンド分析方法およびトレンド分析プログラム
CN115525792B (zh) 视频搜索方法、装置、服务器及终端设备

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: 16906954

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: 16906954

Country of ref document: EP

Kind code of ref document: A1