WO2021196541A1 - 用于搜索内容的方法、装置、设备和计算机可读存储介质 - Google Patents

用于搜索内容的方法、装置、设备和计算机可读存储介质 Download PDF

Info

Publication number
WO2021196541A1
WO2021196541A1 PCT/CN2020/117129 CN2020117129W WO2021196541A1 WO 2021196541 A1 WO2021196541 A1 WO 2021196541A1 CN 2020117129 W CN2020117129 W CN 2020117129W WO 2021196541 A1 WO2021196541 A1 WO 2021196541A1
Authority
WO
WIPO (PCT)
Prior art keywords
historical search
search
historical
result
records
Prior art date
Application number
PCT/CN2020/117129
Other languages
English (en)
French (fr)
Inventor
姜富春
陆伟
史利
王锦东
潘平
赵世奇
袁怀文
金慈航
王彬
欧玉龙
Original Assignee
百度在线网络技术（北京）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 百度在线网络技术（北京）有限公司 filed Critical 百度在线网络技术（北京）有限公司
Priority to EP20929634.2A priority Critical patent/EP4113329A4/en
Priority to KR1020227027825A priority patent/KR20220119745A/ko
Priority to JP2022553192A priority patent/JP7451747B2/ja
Priority to US17/914,557 priority patent/US20230147941A1/en
Publication of WO2021196541A1 publication Critical patent/WO2021196541A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2457—Query processing with adaptation to user needs
    • G06F16/24578—Query processing with adaptation to user needs using ranking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2453—Query optimisation
    • G06F16/24534—Query rewriting; Transformation
    • G06F16/24539—Query rewriting; Transformation using cached or materialised query results
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06F16/332—Query formulation
    • G06F16/3325—Reformulation based on results of preceding query
    • G06F16/3326—Reformulation based on results of preceding query using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06F16/3349—Reuse of stored results of previous queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06F16/9536—Search customisation based on social or collaborative filtering

Definitions

  • the embodiments of the present disclosure mainly relate to the field of data processing, and more specifically, to methods, apparatuses, devices, and computer-readable storage media for searching content.
  • search engines In order to solve the difficulty of information retrieval, many search engines have appeared to help users find information. Because the search engine will collect various information from a large number of websites to the local area, and then build various information databases through processing. When users want to find the content, they can easily and quickly get the content they want to find by entering the search content in the search engine. However, there are still many problems that need to be solved in the process of using search engines to find content.
  • a method for searching content includes obtaining multiple historical search records related to multiple historical search requests in response to receiving a search request for a target search term, each historical search record including a historical search term targeted by a corresponding historical search request.
  • the method further includes determining a first historical search record matching the target search term from a plurality of historical search records.
  • the method further includes determining a second historical search record associated with the first historical search record from the plurality of historical search records based on the relationship between the plurality of historical search records.
  • the method further includes determining an expanded result for the target search item based on the search result corresponding to the second historical search record.
  • an apparatus for searching content includes a historical search record obtaining module, configured to obtain multiple historical search records related to multiple historical search requests in response to receiving a search request for a target search item, and each historical search record includes a corresponding historical search
  • the historical search item targeted by the request is configured to determine the first historical search record matching the target search item from a plurality of historical search records
  • the historical search record determination module is configured to be based on multiple historical search records The relationship between the search records, determining a second historical search record associated with the first historical search record from a plurality of historical search records; and an extended result determination module configured to be based on a search corresponding to the second historical search record As a result, the expanded result for the target search item is determined.
  • an electronic device including one or more processors; and a storage device, for storing one or more programs, when one or more programs are used by one or more processors Execution enables one or more processors to implement the method according to the first aspect of the present disclosure.
  • a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the method according to the first aspect of the present disclosure is implemented.
  • FIG. 1 shows a schematic diagram of an example 100 of providing a recommendation result according to a traditional solution
  • FIG. 2 shows a schematic diagram of an environment 200 in which multiple embodiments of the present disclosure can be implemented
  • FIG. 3 shows a flowchart of a method 300 for searching content according to some embodiments of the present disclosure
  • FIG. 4 shows a flowchart of a method 400 for obtaining multiple historical search records according to some embodiments of the present disclosure
  • FIG. 5 shows a flowchart of a method 500 for determining historical search record categories and relationships according to some embodiments of the present disclosure
  • FIG. 6 shows a flowchart of a method 600 for determining the relationship between historical search records according to some embodiments of the present disclosure
  • FIG. 7 shows a block diagram of an apparatus 700 for searching content according to some embodiments of the present disclosure
  • FIG. 8 shows a block diagram of an apparatus 800 for searching content according to some embodiments of the present disclosure.
  • FIG. 9 shows a block diagram of a device 900 capable of implementing various embodiments of the present disclosure.
  • FIG. 1 shows a schematic diagram of an example 100 in which a traditional solution provides recommended search terms. After the user enters "Liu**" in the search engine, two recommendation boxes 102 and 104 are provided. Some recommended search terms are provided in box 102, and some recommended search terms are also provided in box 104.
  • the recommended search items given by the traditional solution cannot directly meet the relevant needs of the user, and the user is required to click the search item to manually filter the document resources that can meet the needs in the new search page.
  • the search term text in the traditional solution is generally short, and its attractiveness as recommended content is weak, and the search term is generated through user-generated content, and it is difficult to control its quality and safety.
  • an improved solution for searching content is proposed.
  • multiple historical search records related to multiple historical search requests are first obtained, wherein each historical search record includes the corresponding history for which the historical search request is targeted. Search item. Then, the first historical search record matching the target search item is determined from the multiple historical search records. Based on the relationship between the multiple historical search records, a second historical search record associated with the first historical search record is determined from the multiple historical search records. Then, based on the search result corresponding to the second historical search record, the expanded result for the target search item is determined.
  • Figure 2 shows a schematic diagram of an environment 200 in which multiple embodiments of the present disclosure can be implemented.
  • a terminal device 204 and a computing device 208 are included in this example environment 200.
  • the computing device 208 provides the user 202 with an expanded result 212 for the search request 206 based on the search request 206 from the terminal device 204.
  • the terminal device 204 may run an application or program for searching, such as a search engine application.
  • the terminal device 204 receives the target search item input by the user 202, for example, the user 202 inputs "How much is the Mercedes-Benz C200".
  • the terminal device 204 then generates a search request 206 for the target search term and sends the search request 206 to the computing device 208.
  • Terminal devices 204 include, but are not limited to, personal computers, server computers, handheld or laptop devices, mobile devices (such as mobile phones, personal digital assistants (PDAs), media players, etc.), multi-processor systems, consumer electronics, small Computers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.
  • mobile devices such as mobile phones, personal digital assistants (PDAs), media players, etc.
  • PDAs personal digital assistants
  • multi-processor systems consumer electronics
  • small Computers mainframe computers
  • distributed computing environments including any of the above systems or devices, etc.
  • Computing devices 208 include, but are not limited to, personal computers, server computers, handheld or laptop devices, multi-processor systems, consumer electronics, minicomputers, large computers, distributed computing environments including any of the above systems or devices, Virtual machines or other computing devices in the cloud platform.
  • the computing device 208 After the computing device 208 receives the search request 206 from the terminal device 204, the computing device 208 not only generates search results for the target search item in the search request 206, but also obtains extensions from the computing device 208 according to the target search item in the search request 206 Result 212.
  • the multiple historical search records 210 obtained by the computing device 208 are searched for matching historical search records by matching the target search item with the historical search items in the multiple historical search records 210.
  • FIG. 2 shows that the computing device 208 receives a plurality of historical search records 210 from other devices, which is only an example and not a specific limitation of the present disclosure.
  • the plurality of historical search records 210 may also be generated within the computing device 208 or by the computing device 208 when the search request 206 is received.
  • the multiple historical search records 210 are determined by log data in the search log.
  • Each historical search record in the plurality of historical search records 210 includes a historical search item targeted by a corresponding historical search request.
  • each historical search item further includes a key entity, which is performed by entity recognition of the historical search item in the log data, and the number of occurrences of the entity in the historical search item is determined from a plurality of recognized entities. To determine it.
  • each historical search item further includes a category of demand corresponding to the historical search item.
  • each historical search record in addition to the historical search term, each historical search record also includes a historical search term associated with the historical search term and a degree of association with the associated historical search term.
  • the computing device 208 searches the multiple historical search records 210 for historical search terms that are the same as the target search term, for example, searches for historical search records where the historical search term is "How much is a Mercedes-Benz C200?" In some embodiments, the computing device 208 searches the plurality of historical search records 210 for historical search terms whose matching degree with the target search term is higher than a threshold degree.
  • the above examples are only used to describe the present disclosure, but not to specifically limit the present disclosure.
  • the computing device 208 finds the first historical search record that matches the target search term, the computing device 208 also obtains the relationship between the multiple historical search records 210. The computing device 208 then determines the second historical search record associated with the first historical search record based on the relationship between the multiple histories, for example, the historical search item in the second historical search record is "Photo of Mercedes-Benz C200". Alternatively or additionally, the computing device 208 may also determine one or more other historical search records. In some embodiments, the relationship between the multiple historical search records is the degree of correlation with the multiple categories of the multiple historical search records. In some embodiments, the relationship between the multiple historical search records is the degree of association between the multiple historical search records.
  • the computing device 208 then obtains the expanded result 212 based on the historical search terms of the second historical search record. The computing device 208 then provides the expanded result 212 and/or the target search result obtained from the target search term to the user 202.
  • Figure 2 above shows a schematic diagram of an environment 200 in which multiple embodiments of the present disclosure can be implemented.
  • the following describes a flowchart of a method 300 for searching content according to some embodiments of the present disclosure in conjunction with FIG. 3.
  • the method 300 may be implemented by the computing device 208 in FIG. 2 or any other suitable device.
  • the computing device 208 determines whether a search request 206 for the target search term is received. Upon receiving the search request 206, at block 304, the computing device 208 obtains a plurality of historical search records 210 related to the plurality of historical search requests. Wherein, each historical search record includes a historical search item targeted by a corresponding historical search request.
  • each historical search record of the plurality of historical search records 210 includes a historical search term. In some embodiments, each historical search record in the plurality of historical search records 210 includes a historical search term and a key entity corresponding to the historical search term. In some embodiments, each historical search record of the plurality of historical search records 210 includes historical search terms, key entities corresponding to the historical search terms, and corresponding demand categories. In some embodiments, each historical search record of the plurality of historical search records 210 includes a historical search term, a corresponding historical search term, and a degree of association between the historical search term and the corresponding historical search term. The above examples are only used to describe the present disclosure, but not to specifically limit the present disclosure.
  • the plurality of historical search records 210 are obtained by the computing device 208 and other servers or computers. In some embodiments, multiple historical search records 210 have been generated in the computing device 208. In some embodiments, the plurality of historical search records 210 are generated online by the computing device 208 when the user 202 performs a search. The process of obtaining multiple historical search records 210 by the computing device 208 will be described with reference to FIG. 4.
  • the computing device 208 determines the first historical search record matching the target search term from the plurality of historical search records 210. After the computing device 208 obtains the multiple historical search records 210 and the target search term, it will search for the first historical search record matching the multiple historical search records 210 from the multiple historical search records 210.
  • the target search term is exactly the same as the historical search term in the first historical search record.
  • the degree of matching between the target search term and the historical search term in the first historical search record is higher than a predetermined matching threshold.
  • the computing device 208 determines a second historical search record associated with the first historical search record from the plurality of historical search records 210 based on the relationship between the plurality of historical search records 210. In some embodiments, in addition to obtaining the second historical search record, the computing device 208 also obtains other historical search records associated with the first historical search record.
  • each historical search record in the plurality of historical search records 210 includes historical search terms and key entities, or each historical search record includes historical search terms and key entities and the category of each historical search record.
  • the relationship between multiple historical search records 210 is the degree of association between multiple categories.
  • the computing device 208 determines a second category associated with the first category of the first historical search record based on the relationship between the plurality of historical search records 210.
  • the computing device 208 determines from the plurality of historical search records 210 a second historical search record of the second category, the second historical search record including the key entity of the first historical search record.
  • the computing device 208 For each category of the historical search record, the computing device 208 uses the plurality of historical search items included in the plurality of historical search records 210 to determine the category of the plurality of historical search records 210. The computing device 208 then determines the relationship between the plurality of historical search records 210 based on the category. Through the above method, the relationship between the category and multiple historical search records can be determined more quickly and accurately. The process of determining the category and determining the relationship between multiple historical search records related to the category will be described later in conjunction with FIG. 5.
  • the relationship between the plurality of historical search records 210 is obtained. This relationship describes the degree of association between each historical search record in the plurality of historical search records 210 and its corresponding historical search record.
  • the computing device 208 may determine a set of historical search records associated with the first historical search record based on the relationship between the plurality of historical search records 210, and each historical search record in the first historical search record and the set of historical search records Have a degree of relevance.
  • the computing device 208 determines a second historical search record from a set of historical search records based on the degree of association. With this method, the second historical search record with a high matching degree can be found quickly and accurately.
  • the process of determining the degree of association between each historical search record and its corresponding historical search record will be described below in conjunction with FIG. 6.
  • the computing device 208 determines an expanded result 212 for the target search term based on the search result corresponding to the second historical search record.
  • the computing device 208 after obtaining the second historical search record, obtains search results for historical search terms in the second historical search record. In some embodiments, the computing device 208 uses the historical search terms in the second historical search record to re-search, so as to obtain the search result in real time. As an alternative, in some embodiments, the computing device 208 may also look up historical search results regarding historical search terms in the second historical search record. For example, the computing device 208 can look up the aforementioned historical search results from the search log. It should be understood that the above examples are only used to describe the present disclosure, not to specifically limit the present disclosure. The computing device 208 may obtain the search results for the historical search items in the second historical search record in a variety of ways.
  • the computing device 208 determines the search result obtained from the search term in the second historical search record as the expanded result 212. In this way, information suitable for users can be quickly and automatically expanded.
  • the computing device 208 uses the second historical search record to perform a search to obtain historical search results for historical search terms in the second historical search record. For example, the computing device 208 can look up the historical search result from the daily increase in search logs. Then, the computing device 208 determines from the historical search results a part of the historical search results that have been accessed by the user 202. At this time, the computing device 208 determines part of the historical search result as the expanded result 212. In this way, the extension result 212 related to the user can be determined more accurately.
  • the computing device 208 after obtaining the second historical search record, also obtains the information flow generated when the user 202 searches for historical search terms in the second historical search record.
  • the information stream is a historical information stream recorded in the log record that is provided to the user when the user uses the historical search item in the second historical search record to search. Historical information flow can be news, various network information, push advertisements, etc.
  • the computing device 208 determines the information flow browsed by the user 202 during the search as the expanded result 212 based on the information flow. For example, if the user 202 who performed the second historical search record also viewed the information stream pushed from the web server when searching for information, the viewed information stream is used as the extended result 212. Alternatively or additionally, the attention tag established by the user 202 needs to be present in the information stream being viewed. In this way, the sources of extended results can be increased and more extended results can be provided.
  • the computing device 208 may provide the expanded result 212 to the terminal device 204, or the computing device 208 may provide the expanded result 212 and the target search result for the target search term to the terminal device 204. In this way, users can quickly obtain expanded results and target search results.
  • the computing device 208 determines the first score of the expanded result 212, and the first score indicates the expanded result 212 and the historical search in the second historical search record.
  • the score is generated by a neural network model.
  • the score of each result is determined by inputting information such as the user click distribution, user click rate estimation, title, content, length, and historical search items of the second historical search record of each result in the expanded result 212 to the neural network module.
  • the neural network model is determined by sample user click distribution, sample user click rate estimation, sample search result items, sample search items, title, content, length and other information of the expanded results, and sample scores.
  • the computing device 208 also determines a second score of the target search result, the second score indicating the degree of relevance between the target search result and the target search term. It is also inputting the title, content length, target search item and other information of each result of the target search result into the above neural network model to determine the score of the target search result.
  • the computing device 208 determines the priority of the expanded result 212 and the target search result based on the first score and the second score. Then, the computing device 208 provides the expanded result 212 and the target search result according to the priority. Alternatively or as an attachment, the computing device 208 may also set some restriction conditions on the display of the expanded result 212. For example, there can be only the first number of extended results 212 among the predetermined number of results provided, or the continuous number of extended results 212 can be set. The above examples are only used to describe the present disclosure, but not to specifically limit the present disclosure. Those skilled in the art can set it as needed. Through the above method, it is possible to provide users with higher and more accurate target search results and recommendation results.
  • the computing device 208 also establishes a target data source for obtaining search results corresponding to the second historical search record.
  • the target data source may be generated by another device, and then the computing device 208 obtains the target data source from the other device. By establishing the target data source, the quality of the target data source can be improved, so that high-quality content can be provided to users.
  • the computing device 208 when the computing device 208 establishes the target data source, it first determines the scores of multiple documents in the multiple original data sources, and the score of each document indicates the quality of the document.
  • the scoring of documents is determined by the following methods: media site scoring: includes site scoring based on automatic link analysis methods, and site scoring marked by experts; media author scoring: includes author registration marked by experts, author popularity through big data analysis, The popularity of the author synthesized through reader feedback information such as likes and comments; and the richness of media texts, pictures, and videos.
  • the computing device 208 determines the document whose score exceeds the threshold score among the multiple documents as the document in the target data source. In this way, high-quality candidate results can be obtained through truncation operations.
  • FIG. 4 shows a flowchart of a method 400 for obtaining multiple historical search records according to some embodiments of the present disclosure.
  • the method 400 in FIG. 4 may be executed by the computing device 208 in FIG. 2 or any other suitable device.
  • the computing device 208 determines from the search log a set of historical search terms for which a set of historical search requests are targeted.
  • search log entries of all users are stored in the search log. Therefore, a set of historical search terms can be determined from the search log.
  • the computing device 208 determines multiple entities from a set of historical search terms, each entity identifying an object associated with a corresponding historical search term.
  • the computing device 208 performs entity recognition on each historical search item in a set of historical search items, for example, to identify the entity through a named entity recognition method.
  • the computing device 208 determines key entities from the multiple entities based on the number of occurrences of the multiple entities in a set of historical search terms.
  • the computing device 208 determines a set of historical search terms that includes a single entity from a set of historical search terms. Then, the computing device 208 determines at least one historical search item from the set of historical search items, and the number of occurrences of a single entity included in the at least one historical search item in the set of historical search items exceeds the first threshold number of times. The computing device 208 determines a single entity included in at least one historical search term as a key entity. Through this method, key entities can be quickly and accurately determined.
  • the computing device 208 when determining a key entity, determines a high-frequency entity whose number of occurrences exceeds the second threshold number from the multiple entities based on the number of occurrences of the multiple entities in a set of historical search terms. The computing device 208 determines the high-frequency entity as a key entity based on determining that the weight of the high-frequency entity in the corresponding historical search item exceeds the threshold weight, where the weight indicates the importance of the high-frequency entity in the corresponding historical search item.
  • the computing device 208 determines the weight based on the location of the high-frequency entity in the corresponding historical search term. In some embodiments, the computing device 208 determines the weight based on the relationship between the length of the high-frequency entity and the length of the corresponding historical search term. In some embodiments, the computing device 208 may also determine the weight based on the combination of the foregoing methods and using any other suitable information.
  • the above examples are only used to describe the present disclosure, but not to specifically limit the present disclosure. The weights can also be obtained by combining the above methods or in other ways. Through the above method, the weight can be determined accurately and quickly.
  • the computing device 208 selects multiple historical search terms that include key entities from a set of historical search terms. After determining the key entity, the computing device 208 uses the key entity to determine a historical search item that only includes the key entity.
  • the computing device 208 generates a plurality of historical search records 210 based on the plurality of historical search terms and key entities.
  • each historical search record in the plurality of historical search records 210 includes at least a historical search item and its corresponding key entity.
  • the aforementioned multiple historical search records 210 may be generated by other devices based on the search logs, and the computing device 208 receives multiple historical search records 210 from other devices.
  • FIG. 5 shows a flowchart for determining historical search record categories and relationships 500 according to some embodiments of the present disclosure.
  • the method 500 in FIG. 5 may be executed by the computing device 208 in FIG. 2 or any other suitable device.
  • each historical search record in the plurality of historical search records 210 includes key entities in addition to historical search terms.
  • the computing device 208 obtains the remaining parts of each of the multiple historical search items by removing the corresponding key entities from the multiple historical search items. For example, when multiple historical search items are "How much is Mercedes-Benz C200”, “Price of Mercedes-Benz C200”, and “Picture of Mercedes-Benz C200", and the key entity is "Benz C200”, the remaining part is "How much” and " Price”, “Picture of”.
  • the computing device 208 determines demand information associated with a plurality of historical search terms based on at least the remaining portion.
  • the computing device 208 determines the user's demand information for the remaining part, for example, determines the remaining part "how much", “price”, and "picture of" as the demand information.
  • the computing device 208 determines the category of the plurality of historical search records 210 based on the demand information.
  • the computing device 208 uses a clustering operation to process the demand information to determine the categories of the multiple historical search records 210, for example, a k-means method is used to process the demand information.
  • the computing device 208 may also determine the category of the demand information in other suitable ways, such as manually classifying. The above examples are only used to describe the present disclosure, but not to specifically limit the present disclosure.
  • the demand categories of multiple historical search items can be accurately determined, and the classification of multiple historical search records can also be realized.
  • the computing device 208 determines search times or search results for multiple historical search terms from the search log. After determining each category, the computing device 208 needs to determine the association relationship between each category. Therefore, the computing device 208 will determine the log records for multiple historical search items from the search log, and then determine the search time and search results of these log records.
  • the computing device 208 determines the degree of relevance between the multiple categories based on the search time or the search results.
  • the computing device 208 determines in the log that the two historical search records of different categories of the same user within a predetermined period of time are the correlation degree between the two categories increased by 1, alternatively or additionally, the history of the two historical search records
  • the key entities of the search terms are the same. For example, if the user 202 searches for “price of Mercedes-Benz C200” and “picture of Mercedes-Benz C200” within a predetermined time period, it can be determined that the correlation between the category corresponding to “price” and the category corresponding to “picture” is 1.
  • the degree of correlation between multiple categories can be determined.
  • the computing device 208 determines the relationship between the plurality of historical search records 210 based on the degree of correlation between the plurality of categories. Through the degree of correlation between each category, the relationship between multiple historical search records 210 can be determined. For example, when a historical search record has a first category, one or more other categories with a higher degree of relevance can be determined through the first category, and then the key entity of the first historical search record is combined with one or more other categories. The category can then identify other historical search records associated with the historical search record.
  • the degree of correlation between multiple categories can be quickly and accurately determined, so that the accuracy of the recommended results can be ensured when searching.
  • FIG. 6 shows a flowchart for determining a relationship 600 between historical search records according to some embodiments of the present disclosure.
  • the method 600 in FIG. 6 can be executed by the computing device 208 in FIG. 2 or any other suitable device.
  • the computing device 208 determines search times or search results for multiple historical search terms from the search log.
  • the search log stores many search log items of users, and the search time and search results of multiple historical search items can be determined through the search log items.
  • the computing device 208 determines the degree of association between the plurality of historical search records 210 based on the search time or the search result.
  • the computing device 208 determines that there is a correlation between the two search records based on the same user executing two search items within a predetermined period of time or the search results of the two search items having the same result item. For example, if the user 202 executes two historical search items within a predetermined period of time, the degree of association between two historical search records including the two search items may be increased by one. If there are a predetermined number of identical result items in the search results corresponding to the two historical search items, the degree of association between the two historical search records can be increased by one.
  • the computing device 208 determines the relationship between the plurality of historical search records 210 based on the degree of association between the plurality of historical search records 210.
  • the computing device 208 determines the association relationship between the plurality of historical search records 210 based on the determined degree of association.
  • the multiple historical search records 210 and the association relationship between the multiple historical search records 210 may be generated by other devices, and the computing device 208 may be obtained from other devices.
  • association relationship between multiple historical search records can be quickly and accurately determined, so that the expansion result can be quickly and accurately determined.
  • the device 700 includes a high-quality result screening module 702, a related demand mining module 704, a recommendation result matching module 706, and a search result and recommendation result mixing module 708.
  • the relevant demand mining module 704 digs out relevant demands based on the original search terms.
  • the specific expression form of the relevant demands may be in the form of search terms, keyword combinations, semantic vectors, and the like.
  • the recommendation result matching module 706 retrieves the results filtered by the high-quality result screening module 702, and finds resources that can meet the related needs as the recommendation result; finally, the search result and recommendation result mixing module 708 combines the recommendation result with the search
  • the normal results retrieved by the engine are shuffled to form the final result list, which is returned to the user.
  • the related demand mining module 704 mines related needs based on the user's original search terms.
  • the technical methods used include the following: Content-based mining method: First, the search item content is split, and two concepts are defined: search key Entity and demand dimensions.
  • the search core subject is the subject string that the user can extract from the search sequence during the search process. This subject string can represent the user’s core aspirations. For example, the search term is "How much is the Mercedes-Benz c200", the core subject is "Benz c200”, and "How much” is a description of the user's demand for the core subject, where the demand is asking the price.
  • NER named entity recognition
  • the key entity meets three conditions: 1) The key entity itself is used as a search Items have a high number; 2) Key entities frequently appear as substrings in multiple historical search items. 3) Among all the search terms that contain the subject string, the average weight of the subject string is relatively high.
  • the demand dimension is the attribute of the key product entity.
  • the remaining string of the key entity will be removed as the demand.
  • the different expressions of the initially obtained demand substrings may be the same demand, for example, the demand for "How much is the Mercedes-Benz c200" and the "Mercedes-Benz c200 price” is the same.
  • the correlation matrix between different dimensions is calculated to represent the close relationship between the dimensions.
  • Historical search items are separated by key entities and demand information, and by mining a good demand category correlation matrix, the search items with strong correlation demands are regarded as the relevant extended demand collection of the current search items.
  • the graph-based mining method mines the set of search terms that are strongly related to the current search term as the related expansion requirements of the current search term.
  • the core keywords are extracted, and the user's focus tag is established.
  • the recommendation result matching module 706 is based on the relevant requirements mined, and matches the results from the resource library that can meet the relevant requirements.
  • the technical methods used include the following: Matching based on the search and retrieval system: Use the extended search term to search the retrieval system to obtain Satisfied results that match the extended search term. And merge all the results according to the strength of the association as the recommended result of the search term.
  • Matching based on user search big data According to user behaviors such as co-occurrence and a bit, mining articles related to the extended search term as the recommendation result of the search term.
  • Matching of big data based on user search and information flow browsing Mining and counting user focus tags from user search and information flow browsing data, and recalling articles through focus matching, as the user's personalized recommendation results.
  • the target search result and extended result mixing module 708 mainly includes search result scoring, recommendation result scoring, and mixing. Scoring of search results: scoring based on the fusion model of historical click distribution and user click rate estimation. The recommendation result score is based on historical click distribution, user click-through rate estimation scoring, etc.
  • shuffle sort from high to low based on search result score and recommendation result score.
  • diversity control will also be carried out, including diversity control based on the density of recommended results, and diversity control of the density of recommended results on the same theme.
  • the high-quality result screening module 702 scores the document resources based on some basic quality factors, and truncates based on the scores to obtain high-quality candidate results.
  • the basic quality factors include: media site scoring: including site scoring based on automatic link analysis methods and site scoring marked by experts; media author scoring: including author registration marked by experts, author popularity through big data analysis, through likes, and comments The author’s popularity based on reader feedback information; the richness of media texts, pictures, and videos.
  • FIG. 8 shows a schematic block diagram of an apparatus 800 for searching content according to an embodiment of the present disclosure.
  • the apparatus 800 may include a historical search record obtaining module 802, configured to obtain multiple historical search records related to multiple historical search requests in response to receiving a search request for a target search item, each historical search record The search record includes the historical search item targeted by the corresponding historical search request.
  • the device 800 further includes a target search term matching module 804 configured to determine a first historical search record matching the target search term from a plurality of historical search records.
  • the device 800 further includes a historical search record determination module 806 configured to determine a second historical search record associated with the first historical search record from the multiple historical search records based on the relationship between the multiple historical search records.
  • the device 800 further includes an extended result determination module configured to determine an extended result for the target search item based on the search result corresponding to the second historical search record.
  • the historical search record acquisition module 802 includes: a first historical search item determination module configured to determine a set of historical search items targeted by a set of historical search requests from the search log; an entity determination module configured to To determine multiple entities from a set of historical search terms, each entity identifies an object associated with a corresponding historical search term; the first key entity determination module is configured to be based on the appearance of multiple entities in a set of historical search terms The number of times, the key entity is determined from multiple entities; the selection module is configured to select multiple historical search items including key entities from a set of historical search items; and the generation module is configured to be based on multiple historical search items and key The entity generates multiple historical search records.
  • the first key entity determining module includes a historical search item set determining module configured to determine a historical search item set including a single entity from a set of historical search items; the second historical search item determining module is configured In order to determine at least one historical search item from the set of historical search items, the number of occurrences of a single entity included in the at least one historical search item in the set of historical search items exceeds a first threshold number of times; and a key entity determining module for a single entity is configured to A single entity included in at least one historical search item is determined as a key entity.
  • the key entity determination module includes a high-frequency entity determination module configured to determine from a plurality of entities that the number of occurrences exceeds a second threshold, based on the number of occurrences of multiple entities in a set of historical search terms. Frequency entity; and a second key entity determination module, configured to determine the high frequency entity as a key entity based on determining that the weight of the high frequency entity in the corresponding historical search item exceeds the threshold weight, wherein the weight indicates that the high frequency entity is in the corresponding Importance in historical search terms.
  • the second key entity determination module includes a position determination module configured to determine the position of the high-frequency entity in the corresponding historical search item, and a length relationship determination module configured to determine the length of the high-frequency entity and the corresponding The relationship between the length of historical search terms.
  • the device 800 further includes a category determining module configured to determine the categories of multiple historical search records based on multiple historical search items included in the multiple historical search records; and a historical search record relationship determining module configured to To determine the relationship between multiple historical search records based on categories.
  • each historical search record in the plurality of historical search records further includes a key entity
  • the category determination module includes: a remaining part determination module configured to remove the corresponding key entity from the plurality of historical search items , Obtain the remaining parts of each of the multiple historical search items; a requirement information determination module configured to determine the requirement information associated with the multiple historical search items based at least on the remaining parts; and the historical search record category determining module, configured to be based on requirements Information to determine the category of multiple historical search records.
  • the multiple historical search records have multiple categories
  • the historical search record relationship determination module includes: a first search time or search result determination module configured to determine from the search log for multiple historical search terms The search time or search results of the search time; the correlation degree determination module is configured to determine the correlation degree between multiple categories based on the search time or search results; and the relationship determination module based on the correlation degree is configured to determine the relationship between multiple categories based on the search time or search results. The degree of relevance to determine the relationship between multiple historical search records.
  • the apparatus 800 further includes a second search time or search result configuration module, configured to determine the search time or search results for multiple historical search items from the search log; the association degree determination module is configured to be based on The search time or search results determine the degree of association between multiple historical search records; and the relationship determination module based on the degree of association is configured to determine the degree of association between the multiple historical search records based on the degree of association between the multiple historical search records. Relationship between.
  • each historical search record in the plurality of historical search records further includes a key entity and a category of each historical search record
  • the historical search record determination module 806 includes: a second category determination module configured to be based on The relationship between a plurality of historical search records is determined to determine a second category associated with the first category of the first historical search record; and a second historical search record determination module having a category is configured to determine from the multiple historical search records There is a second historical search record of a second category, and the second historical search record includes a key entity of the first historical search record.
  • the historical search record determination module 806 includes a set of historical search record determination modules configured to determine a set of historical search records associated with the first historical search record based on the relationship between the multiple historical search records ,
  • the first historical search record has a degree of association with each historical search record in a set of historical search records; and the historical search record determining module based on the degree of association is configured to determine the first historical search record from the set of historical search records based on the degree of association 2.
  • the extended result determination module 808 includes a first search result acquisition module configured to acquire search results for historical search items in the second historical search record; and an extended result determination module for search results is configured To determine the search result as an extended result.
  • the extended result determination module 808 includes a second search result acquisition module configured to acquire historical search results for historical search items in the second historical search record; a partial historical search result determination module is configured to obtain Part of the historical search results that are determined to have been accessed by the user in the historical search results; and a part of the historical search result expansion module configured to determine the part of the historical search results as expanded results.
  • the extended result determination module includes an information flow module configured to obtain the information flow generated when the user searches for the historical search items in the second historical search record; and the extended result module of the targeted information flow , Is configured to determine the extension result based on the information flow.
  • the device 800 further includes at least one of the following: a first providing device configured to provide extended results; and a second providing device configured to provide extended results and target search results for target search terms.
  • the second providing device includes a first score determination module configured to determine a first score of the expanded result, the first score indicating the degree of relevance between the expanded result and the historical search item in the second historical search record;
  • the second score determination module is configured to determine the second score of the target search result, the second score indicates the degree of relevance between the target search result and the target search item;
  • the priority determination module is configured to determine based on the first score and the second score The priority of the expanded result and the target search result; and the expanded result and search result providing module configured to provide the expanded result and the target search result based on the priority.
  • the device 800 further includes a target data source establishment module configured to establish a target data source for obtaining search results corresponding to the second historical search record.
  • the target data source establishment module includes a document score determination module configured to determine the scores of multiple documents in the multiple original data sources, the score of each document indicates the quality of the document; and the target data source document determination The module is configured to determine the document whose score exceeds the threshold score among multiple documents as the document in the target data source.
  • FIG. 9 shows a schematic block diagram of an electronic device 900 that can be used to implement embodiments of the present disclosure.
  • the device 900 may be used to implement the terminal device 204 and the computing device 208 in FIG. 1.
  • the device 900 includes a computing unit 901, which can be configured according to computer program instructions stored in a read-only memory (ROM) 902 or computer program instructions loaded from a storage unit 808 to a random access memory (RAM) 903. Perform various appropriate actions and processing.
  • ROM read-only memory
  • RAM random access memory
  • the computing unit 901, the ROM 902, and the RAM 903 are connected to each other through a bus 904.
  • An input/output (I/O) interface 905 is also connected to the bus 904.
  • the I/O interface 905 includes: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; and a storage unit 908, such as a magnetic disk, an optical disk, etc. ; And the communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc.
  • the communication unit 909 allows the device 900 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
  • the computing unit 901 may be various general-purpose and/or special-purpose processing components with processing and computing capabilities. Some examples of computing unit 901 include, but are not limited to, central processing unit (CPU), graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing DSP, and any appropriate processor, controller, microcontroller, etc.
  • the calculation unit 901 executes the various methods and processes described above, such as methods 300, 400, 500, and 600.
  • 300, 400, 500, and 600 may be implemented as computer software programs, which are tangibly contained in a machine-readable medium, such as the storage unit 908.
  • part or all of the computer program may be loaded and/or installed on the device 900 via the ROM 902 and/or the communication unit 909.
  • the computer program When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the methods 300, 400, 500, and 600 described above can be executed.
  • the computing unit 901 may be configured to execute the method 900 in any other suitable manner (for example, by means of firmware).
  • exemplary types of hardware logic components include: Field Programmable Gate Array (FPGA), Application Specific Integrated Circuit (ASIC), Application Specific Standard Product (ASSP), System on Chip System (SOC), Load programmable logic device (CPLD) and so on.
  • FPGA Field Programmable Gate Array
  • ASIC Application Specific Integrated Circuit
  • ASSP Application Specific Standard Product
  • SOC System on Chip System
  • CPLD Load programmable logic device
  • the program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to the processors or controllers of general-purpose computers, special-purpose computers, or other programmable data processing devices, so that when the program codes are executed by the processor or controller, the functions specified in the flowcharts and/or block diagrams/ The operation is implemented.
  • the program code can be executed entirely on the machine, partly executed on the machine, partly executed on the machine and partly executed on the remote machine as an independent software package, or entirely executed on the remote machine or server.
  • a machine-readable medium may be a tangible medium, which may contain or store a program for use by the instruction execution system, apparatus, or device or in combination with the instruction execution system, apparatus, or device.
  • the machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • the machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing.
  • machine-readable storage media would include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing.
  • RAM random access memory
  • ROM read-only memory
  • EPROM or flash memory erasable programmable read-only memory
  • CD-ROM compact disk read only memory
  • magnetic storage device or any suitable combination of the foregoing.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种用于搜索内容的方法、装置、设备和计算机可读存储介质，涉及数据处理领域。在该方法中，响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。然后，从多个历史搜索记录中确定与目标搜索项匹配的第一历史搜索记录。进而，基于多个历史搜索记录之间的关系，从多个历史搜索记录中确定与第一历史搜索记录相关联的第二历史搜索记录。该方法还包括基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果。通过该方法，能够提供可能满足用户搜索需求的扩展结果，提高了搜索质量和效果，改进了用户体验。

Description

用于搜索内容的方法、装置、设备和计算机可读存储介质 技术领域
本公开的实施例主要涉及数据处理领域，并且更具体地，涉及用于搜索内容的方法、装置、设备和计算机可读存储介质。
背景技术
随着信息技术的快速发展，很多用户和网站提供了大量的信息可供访问。然后，由于网站提供的数据量越来越大，对于单个用户而言，要去各个网站或信息源查找相关的信息变得非常困难。
为了解决信息检索的困难，出现了许多搜索引擎来帮助用户查找信息。由于搜索引擎会从大量的网站的将各种信息收集到本地，然后经过加工建立各种信息数据库。当用户想要查找的内容时，通过在搜索引擎输入搜索内容便可以轻松快速的获得想要查找的内容。然而，在使用搜索引擎查找内容的过程中还存在着许多需要解决的问题。
发明内容
根据本公开的示例实施例，提供了一种用于搜索内容的方案。
在本公开的第一方面中，提供了一种用于搜索内容的方法。该方法包括响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。该方法还包括从多个历史搜索记录中确定与目标搜索项匹配的第一历史搜索记录。该方法还包括基于多个历史搜索记录之间的关系，从多个历史搜索记录中确定与第一历史搜索记录相关联的第二历史搜索记录。该方法还包括基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果。
在本公开的第二方面中，提供了一种搜索内容的装置。该装置包括历史搜索记录获取模块，被配置为响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项；目标搜索项匹配模块，被配置为从多个历史搜索记录中确定与目标搜索项匹配的第一历史搜索记录；历史搜索记录确定模块，被配置为基于多个历史搜索记录之间的关系，从多个历史搜索记录中确定与第一历史搜索记录相关联的第二历史搜索记录；以及扩展结果确定模块，被配置为基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果。
在本公开的第三方面中，提供了一种电子设备，包括一个或多个处理器；以及存储装置，用于存储一个或多个程序，当一个或多个程序被一个或多个处理器执行，使得一个或多个处理器实现根据本公开的第一方面的方法。
在本公开的第四方面中，提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现根据本公开的第一方面的方法。
附图说明
结合附图并参考以下详细说明，本公开各实施例的上述和其他特征、优点及方面将变得更加明显。在附图中，相同或相似的附图标注表示相同或相似的元素，其中：
图1示出了根据传统方案来提供推荐结果的示例100的示意图；
图2示出了本公开的多个实施例的能够在其中实现的环境200的示意图；
图3示出了根据本公开的一些实施例的用于搜索内容的方法300的流程图；
图4示出了根据本公开的一些实施例的用于获取多个历史搜索记录的方法400的流程图；
图5示出了根据本公开的一些实施例的用于确定历史搜索记录类别和关系的方法500的流程图；
图6示出了根据本公开的一些实施例的用于确定历史搜索记录间的关系的方法600的流程图；
图7示出了根据本公开的一些实施例的用于搜索内容的装置700的框图；
图8示出了根据本公开的一些实施例的用于搜索内容的装置800的框图；以及
图9示出了能够实施本公开的多个实施例的设备900的框图。
具体实施方式
下面将参照附图更详细地描述本公开的实施例。虽然附图中显示了本公开的某些实施例，然而应当理解的是，本公开可以通过各种形式来实现，而且不应该被解释为限于这里阐述的实施例，相反提供这些实施例是为了更加透彻和完整地理解本公开。应当理解的是，本公开的附图及实施例仅用于示例性作用，并非用于限制本公开的保护范围。
在本公开的实施例的描述中，术语“包括”及其类似用语应当理解为开放性包含，即“包括但不限于”。术语“基于”应当理解为“至少部分地基于”。术语“一个实施例”或“该实施例”应当理解为“至少一个实施例”。术语“第一”、“第二”等等可以指代不同的或相同的对象。下文还可能包括其他明确的和隐含的定义。
在搜索引擎中，用户提交搜索项，搜索引擎对网页库进行检索。然后搜索引擎获取和搜索项匹配的结果，对结果进行排序后返回给用户。除了搜索项所明确表达的信息需求之外，用户往往同时有一些相关的信息需求。因此在传统的搜索引擎产品中，都具有一定的推荐功能，为当前的搜索项推荐相关的搜索项，以满足用户这些相关的信息需求。例如，图1示出了传统方案提供推荐的搜索项的示例100的示意图。在用户在搜索引擎中输入“刘**”后提供了两个 推荐框102和104。在框102中提供了一些推荐搜索项，而在框104中也提供了一些推荐搜索项。
然而，传统方案给出的推荐搜索项无法直接满足用户的相关需求，并且需要用户点击搜索项，在新搜索页面中人工筛选可以满足需求的文档资源。此外，传统方案中的搜索项文本普遍较短，其作为推荐内容的吸引力弱，并且搜索项为通过用户生成内容的方式产生，难以控制其质量和安全性。
根据本公开的实施例，提出一种搜索内容的改进方案。在该方案中，在接收到针对目标搜索项的搜索请求时，先获取与多个历史搜索请求有关的多个历史搜索记录，其中每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。然后从多个历史搜索记录中确定与目标搜索项匹配的第一历史搜索记录。通过多个历史搜索记录之间的关系，从多个历史搜索记录中确定与第一历史搜索记录相关联的第二历史搜索记录。然后基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果。通过该方法，能够提供可能满足用户搜索需求的扩展结果，提高了搜索质量和效果，改进了用户体验。
图2示出了本公开的多个实施例能够在其中实现的环境200的示意图。在该示例环境200中，在该示例环境200中包括终端设备204和计算设备208。计算设备208基于来自终端设备204的搜索请求206来为用户202提供与针对搜索请求206的扩展结果212。
终端设备204可以运行用于搜索的应用或程序，诸如搜索引擎应用。终端设备204接收用户202输入的目标搜索项，例如用户202输入“奔驰C200多少钱”。然后终端设备204生成针对该目标搜索项的搜索请求206并将搜索请求206发送到计算设备208。
终端设备204包括但不限于个人计算机、服务器计算机、手持或膝上型设备、移动设备(诸如移动电话、个人数字助理(PDA)、媒体播放器等)、多处理器系统、消费电子产品、小型计算机、大型计算机、包括上述系统或设备中的任意一个的分布式计算环境等。
计算设备208包括但不限于个人计算机、服务器计算机、手持或膝上型设备、多处理器系统、消费电子产品、小型计算机、大型计算机、包括上述系统或设备中的任意一个的分布式计算环境、云平台中的虚拟机或其他计算设备等。
计算设备208在接收到来自终端设备204的搜索请求206后，计算设备208不仅生成针对搜索请求206中的目标搜索项的搜索结果，还会根据搜索请求206的目标搜索项从计算设备208获得扩展结果212。计算设备208获得的多个历史搜索记录210，通过将目标搜索项与多个历史搜索记录210中的历史搜索项进行匹配查找匹配的历史搜索记录。
图2中示出了计算设备208从其他设备接收多个历史搜索记录210，其仅是示例，而非对本公开的具体限定。多个历史搜索记录210也可以在计算设备208内或由计算设备208在接收到搜索请求206时生成。
多个历史搜索记录210是由搜索日志中的日志数据确定的。多个历史搜索记录210中的每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。在一些实施例中，每个历史搜索项还包括关键实体，该关键实体是通过对日志数据中的历史搜索项进行实体识别，从多个识别的实体中根据实体在历史搜索项中出现的次数来确定出来的。备选地或附加地，每个历史搜索项还包括与该历史搜索项相对应的需求的类别。在一些实施例中，每个历史搜索记录除了历史搜索项之外，还包括与该历史搜索项相关联的历史搜索项，以及与相关联的历史搜索项的关联程度。
在一些实施例中，计算设备208在多个历史搜索记录210中查找与目标搜索项相同的历史搜索项，例如查找历史搜索项为“奔驰C200多少钱”的历史搜索记录。在一些实施例中，计算设备208在多个历史搜索记录210查找与目标搜索项匹配程度高于阈值程度的历史搜索项。上述示例仅是用于描述本公开，而非对本公开的具体限定。
计算设备208在查找到与目标搜索项相匹配的第一历史搜索记录时，计算设备208还会获得多个历史搜索记录210之间的关系。然后计算设备208基于多个历史之间的关系来确定出与第一历史搜索记录相关联的第二历史搜索记录，例如第二历史搜索记录中的历史搜索项为“奔驰C200的照片”。备选地或附加地，计算设备208还可以确定出其他的一个或多个历史搜索记录。在一些实施例中，多个历史搜索记录之间的关系是与多个历史搜索记录的多个类别之间的相关程度。在一些实施例中，多个历史搜索记录之间的关系是多个历史搜索记录之间的关联程度。
计算设备208然后基于第二历史搜索记录的历史搜索项来获得扩展结果212。然后计算设备208将扩展结果212和/或由目标搜索项获得的目标搜索结果提供给用户202。
上面图2示出了本公开的多个实施例能够在其中实现的环境200的示意图。下面结合图3描述根据本公开的一些实施例的用于搜索内容的方法300的流程图。方法300可以由图2中的计算设备208或其它任意合适的设备来实现。
在框302处，计算设备208会确定是否接收到针对目标搜索项的搜索请求206。在接收到搜索请求206时，在框304处，计算设备208获取与多个历史搜索请求有关的多个历史搜索记录210。其中，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。
在一些实施例中，多个历史搜索记录210的每个历史搜索记录包括历史搜索项。在一些实施例中，多个历史搜索记录210中的每个历史搜索记录包括历史搜索项和与历史搜索项相对应的关键实体。在一些实施例中，多个历史搜索记录210的每个历史搜索记录包括历史搜索项与历史搜索项相对应的关键实体及其对应的需求类别。在一些实施例中，多个历史搜索记录210的每个历史搜索记录包括历史搜索项、对应的历史搜索项以及历史搜索项和对应历史搜索项之间的关联程度。上述示例仅是用于描述本公开，而非对本公 开的具体限定。
在一些实施例中，多个历史搜索记录210是计算设备208多其他服务器或计算机获取的。在一些实施例中，多个历史搜索记录210在计算设备208内已生成好。在一些实施例中，多个历史搜索记录210是在用户202进行检索时由计算设备208在线生成的。计算设备208获取多个历史搜索记录210的过程将结合图4进行描述。
在框306处，计算设备208从多个历史搜索记录210中确定与目标搜索项匹配的第一历史搜索记录。计算设备208在获得多个历史搜索记录210以及目标搜索项后，会从多个历史搜索记录210中查找与多个历史搜索记录210相匹配的第一历史搜索记录。在一些实施例中，目标搜索项与第一历史搜索记录中的历史搜索项完全相同。在一些实施例中，目标搜索项与第一历史搜索记录中的历史搜索项的匹配程度高于预定的匹配阈值。上述示例仅是用于描述本公开，而非对本公开的具本限定。
在框308处，计算设备208基于多个历史搜索记录210之间的关系，从多个历史搜索记录210中确定与第一历史搜索记录相关联的第二历史搜索记录。在一些实施例中，计算设备208除了获得第二历史搜索记录之外，还会获取与第一历史搜索记录相关联的其他历史搜索记录。
在一些实施例中，在多个历史搜索记录210中的每个历史搜索记录包括历史搜索项和关键实体、或每个历史搜索记录包括历史搜索项和关键实体和每个历史搜索记录的类别时，多个历史搜索记录210之间的关系是多个类别之间的关联程度。计算设备208基于多个历史搜索记录210之间的关系，确定与第一历史搜索记录的第一类别相关联的第二类别。然后计算设备208从多个历史搜索记录210确定具有第二类别的第二历史搜索记录，第二历史搜索记录包括第一历史搜索记录的关键实体。通过上述方法，可以快速准确的查找到高匹配程度的第二历史搜索记录。
对于每个历史搜索记录的类别，计算设备208利用多个历史搜 索记录210包括的多个历史搜索项来确定多个历史搜索记录210的类别。然后计算设备208基于类别，确定多个历史搜索记录210之间的关系。通过上述方式，可以更快更准确地确定类别和多个历史搜索记录之间的关系。确定类别和确定与类别相关的多个历史搜索记录之间的关系的过程将在后面结合图5进行描述。
在一些实施例中，在计算设备208获取多个历史搜索记录210时会获取多个历史搜索记录210之间的关系。该关系描述了多个历史搜索记录210中的每个历史搜索记录和与其对应的历史搜索记录之间的关联程度。计算设备208可以基于多个历史搜索记录210之间的关系，确定与第一历史搜索记录相关联的一组历史搜索记录，第一历史搜索记录与一组历史搜索记录中的每个历史搜索记录具有关联程度。然后计算设备208基于关联程度，从一组历史搜索记录中确定第二历史搜索记录。通过该方法，可以快速准确的查找到高匹配程度的第二历史搜索记录。确定每个历史搜索记录和与其对应的历史搜索记录之间的关联程度的过程将在下面结合图6进行描述。
在框310处，计算设备208基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果212。
在一些实施例中，在获得第二历史搜索记录之后，计算设备208获取针对第二历史搜索记录中的历史搜索项的搜索结果。在一些实施例中，计算设备208利用第二历史搜索记录中的历史搜索项重新进行搜索，从而实时获得该搜索结果。作为备选方式，在一些实施例中，计算设备208也可以查找关于第二历史搜索记录中的历史搜索项的历史搜索结果。举例而言，计算设备208可以从搜索日志中查找上述历史搜索结果。应当理解，上述示例仅是用于描述本公开，而非对本公开的具体限定，计算设备208可以通过多种方式来获取针对第二历史搜索记录中的历史搜索项的搜索结果。
然后，计算设备208将由第二历史搜索记录中的搜索项得到的搜索结果确定为扩展结果212。通过这种方式，可以快速自动的扩展出适合用户的信息。
在一些实施例中，在获得第二历史搜索记录之后，计算设备208会利用第二历史搜索记录来进行搜索以获取针对第二历史搜索记录中的历史搜索项的历史搜索结果。例如，计算设备208可以从搜索日增收志中查找该历史搜索结果。然后，计算设备208从历史搜索结果中确定已被用户202访问的部分历史搜索结果。此时，计算设备208将部分历史搜索结果确定为扩展结果212。通过这种方式，可以更准确的确定出与用户相关的扩展结果212。
在一些实施例中，在获得第二历史搜索记录之后，计算设备208还会获取用户202在针对第二历史搜索记录中的历史搜索项进行搜索时所产生的信息流。在一些实施例中，该信息流是在日志记录中记录的用户在使用第二历史搜索记录中的历史搜索项进行搜索时，提供给用户的历史信息流。历史信息流可以新闻、各种网络信息、推送广告等。然后计算设备208基于该信息流，将用户202在搜索时浏览的信息流确定为扩展结果212。例如，执行第二历史搜索记录的用户202在搜索信息时还查看了来自网络服务器推送的信息流，则该被查看的信息流作为扩展结果212。备选地或附加地，被查看的信息流中需要存在用户202建立的关注点标签。通过这种方式，可以增加扩展结果的来源，提供更多的扩展结果。
在一些实施例中，在获得扩展结果212后，计算设备208会向终端设备204提供扩展结果212，或者计算设备208会向终端设备204提供扩展结果212和针对目标搜索项的目标搜索结果。通过这种方式，可以使和户快速获得扩展结果和目标搜索结果。
在一些实施例中，在向用户202提供扩展结果212和目标搜索结果时，计算设备208会确定扩展结果212的第一分数，第一分数指示扩展结果212与第二历史搜索记录中的历史搜索项的相关度。该分数是通过神经网络模型来生成的。通过向神经网络模块输入扩展结果212中每项结果的用户点击分布、用户点击率预估、标题、内容、长度和第二历史搜索记录的历史搜索项等信息来确定每项结果的分数。该神经网络模型是通过样本用户点击分布、样本用户点 击率预估样本搜索结果项、样本搜索项、扩展结果的标题、内容、长度等信信息以及样本分数确定的。
计算设备208还会确定目标搜索结果的第二分数，第二分数指示目标搜索结果与目标搜索项的相关度。其也是向上述神经网络模型中输入目标搜索结果的每项结果的标题、内容长度、目标搜索项等信息来确定目标搜索结果的分数。
计算设备208基于第一分数和第二分数，确定扩展结果212和目标搜索结果的优先级。然后，计算设备208根据优先级提供扩展结果212和目标搜索结果。备选地或附中地，计算设备208还可以对扩展结果212的显示设置一些限制条件。如提供的预定数量的结果中只能有第一数量的扩展结果212，或者设置扩展结果212连续的数目等。上述示例仅是用于描述本公开，而非对本公开的具体限定。本领域技术人员可以依据需来设置。通过上述方法，可以向用户相关度更高和更准确的目标搜索结果和推荐结果。
在一些实施例中，计算设备208还会建立用于获得与第二历史搜索记录相对应的搜索结果的目标数据源。在一些实施例中，该目标数据源可以是其他设备生成，然后计算设备208从其他设备获取该目标数据源。通过建立目标数据源，可以提高目标数据源的质量，从而可以为用户提供高质量的内容。
在一些实施例中，计算设备208在建立目标数据源时，先确定多个原始数据源中的多个文档的分数，每个文档的分数指示文档的质量。对文档打分时通过以下方式来确定：媒体站点打分：包括基于自动链接分析方法的站点打分，以及专家标注的站点打分；媒体作者打分：包括专家标注的作者登记、通过大数据分析的作者知名度、通过点赞、评论等读者反馈信息综合出的作者受欢迎度；以及媒体文本、图片、视频的丰富程度。
然后，计算设备208将多个文档中分数超过阈值分数的文档确定为目标数据源中的文档。通过这种方式，可以通过截断操作来获取优质候选结果。
上面结合图3描述了根据本公开的一些实施例的用于搜索内容的方法300的流程图。下面结合图4详细描述图3中的框304处的获取多个历史搜索记录的过程。图4示出了根据本公开的一些实施例的用于获取多个历史搜索记录的方法400的流程图。图4中的方法400可以图2中的计算设备208或其它任意合适的设备来执行。
在框402处，计算设备208从搜索日志中确定一组历史搜索请求所针对的一组历史搜索项。通常，搜索日志中会存储所有用户的搜索日志项。因此，可以从搜索日志中确定出一组历史搜索项。
在框404处，计算设备208从一组历史搜索项中确定多个实体，每个实体标识与对应历史搜索项相关联的对象。计算设备208会对一组历史搜索项中的每个历史搜索项执行实体识别，例如通过命名实体识别方法来识别实体。
在框406处，计算设备208基于多个实体在一组历史搜索项中的出现次数，从多个实体中确定关键实体。
在一些实施例中，计算设备208从一组历史搜索项中确定包括单个实体的历史搜索项集合。然后，计算设备208从历史搜索项集合确定至少一个历史搜索项，至少一个历史搜索项包括的单个实体在历史搜索项集合中的出现次数超过第一阈值次数。计算设备208将至少一个历史搜索项包括的单个实体确定为关键实体。通过这种方法，可以快速准确的确定出关键实体。
例如，假如一组历史搜索项中由实体“奔驰C200”作为历史搜索项的数目为4，而在阈值次数为3时，则可以将“奔驰C200”作为关键实体。
在一些实施例中，在确定关键实体时，计算设备208基于多个实体在一组历史搜索项中的出现次数，从多个实体中确定出现次数超过第二阈值次数的高频实体。计算设备208根据确定高频实体在对应的历史搜索项中的权重超过阈值权重，将高频实体确定为关键实体，其中权重指示高频实体在对应的历史搜索项中的重要性。通过上述方法，可以快速准确的确定出关键实体。
在一些实施例中，计算设备208通过高频实体在对应的历史搜索项中的位置来确定权重。在一些实施例中，计算设备208根据高频实体的长度与对应的历史搜索项的长度之间的关系来确定权重。在一些实施例中，计算设备208还可以根据上述方法的给合以及利用任意其他合适的信息来确定权重。上述示例仅是和于描述本公开，而非对本公开的具体限定，也可以由上述方式进行组合或其他方式来获得权重。通过上述方法，可以准确、快速的确定出权重。
在框408处，计算设备208从一组历史搜索项中选择包括关键实体的多个历史搜索项。在确定出关键实体后，计算设备208利用关键实体来确定出仅包括关键实体的历史搜索项。
在框410处，计算设备208根据多个历史搜索项和关键实体生成多个历史搜索记录210。此时，多个历史搜索记录210中的每个历史搜索记录至少包括历史搜索项和其对应的关键实体。
在一些实施例中，上述多个历史搜索记录210可以由其他设备根据搜索日志生成，计算设备208从其他设备接收多个历史搜索记录210。
通过上述方法，可以从搜索日志中快速准确地确定出包括关键实体的多个历史搜索记录，从而可以使得推荐结果更准确。
上面结合图4描述了根据本公开的一些实施例的用于获取多个历史搜索记录的方法400的流程图。下面结合图5详细描述图3中的框308处的确定历史搜索记录间的类别和关系的过程。图5示出了根据本公开的一些实施例的用于确定历史搜索记录类别和关系500的流程图。图5中的方法500可以用图2中的计算设备208或其它任意合适的设备来执行。
在框502处，多个历史搜索记录210中的每个历史搜索记录除了包括历史搜索项之外，还包括关键实体。计算设备208通过从多个历史搜索项中去除相应的关键实体，获得多个历史搜索项各自的剩余部分。例如，在多个历史搜索项为“奔驰C200多少钱”、“奔驰C200的价格”、“奔驰C200的图片”时，关键实体为“奔驰C200” 时，剩余部分为“多少钱”、“的价格”、“的图片”。
在框504处，计算设备208至少基于剩余部分确定与多个历史搜索项相关联的需求信息。计算设备208将剩余部分确定用户的需求信息，例如将剩余部分“多少钱”、“的价格”、“的图片”确定为需求信息。
在框506处，计算设备208基于需求信息来确定多个历史搜索记录210的类别。在一些实施例中，计算设备208采用聚类操作来处理需求信息以确定多个历史搜索记录210的类别，例如采用k-means方法来处理需求信息。在一些实施例中，计算设备208也可以通过其他合适的方式来确定需求信息的类别，例如通过人工来进行分类。上述示例仅是用于描述本公开，而非对本公开的具体限定。
通过上述方式，可以准确的确定出多个历史搜索项的需求类别，也实现了对多个历史搜索记录的分类。
在框508处，计算设备208从搜索日志中确定针对多个历史搜索项的搜索时间或搜索结果。在确定好各个类别后，计算设备208需要确定各个类别之间的关联关系。因此，计算设备208会再从搜索日志中确定针对多个历史搜索项的目志记录，在后确定出这些日志记录的搜索时间和搜索结果。
在框510处，计算设备208基于搜索时间或搜索结果，确定多个类别之间的相关程度。计算设备208在日志中同一用户在预定时间内的具有不同类别的两次历史搜索记录确定为这两个类别之间的相关程度增加1，备选地或附加地，两次历史搜索记录的历史搜索项的关键实体相同。例中，用户202在预定时段内搜索了“奔驰C200的价格”和“奔驰C200的图片”，则可以确定“价格”对应的类别和“图片”对应的类别之间的相关程度为1。以此类推，可以确定多个类别之间的相关程度。
在框512处，计算设备208基于多个类别之间的相关程度，确定多个历史搜索记录210之间的关系。通过各个类别之间的相关程度，可以确定出多个历史搜索记录210之间的关系。例如，在一个 历史搜索记录具有第一类别时，可以通过第一类别确定出与其相关程度较高的一个或多个其他类别，然后通过该第一历史搜索记录的关键实体结合一个或多个其他类别便能确定出与该历史搜索记录相关联的其他历史搜索记录。
通过上述方法，可以快速准确的确定多个类别之间的相关程度，以使得可以在进行搜索时确保推荐结果的准确性。
上面结合图5描述了根据本公开的一些实施例的用于确定历史搜索记录类别和关系500的流程图。下面结合图6详细描述图3中的框308处的确定历史搜索记录间的关系的过程。图6示出了根据本公开的一些实施例的用于确定历史搜索记录间的关系600的流程图。图6中的方法600可以图2中的计算设备208或其它任意合适的设备来执行。
在框602处，计算设备208从搜索日志中确定针对多个历史搜索项的搜索时间或搜索结果。搜索日志内存储了很多用户的搜索日志项，通过该搜索日志项可以确定出多个历史搜索项的搜索时间和搜索结果。
在框604处，计算设备208基于搜索时间或搜索结果，确定多个历史搜索记录210的之间的关联程度。计算设备208基于同一用户在预定时段内执行两个搜索项或者两个搜索项的搜索结果中具有相同的结果项而确定两个搜索记录之间具有相关性。例如用户202在预定时段内执行两个历史搜索项，则可以将包括两个搜索项的两个历史搜索记录之间的关联程度增加1。如果两个历史搜索项对应的搜索结果中有预定数目的相同的结果项，则可以将两个历史搜索记录之间的关联程度增加1。上述示例仅是用于描述本公开，而非对本公开的具体限定。备选地或附加地，还可以通相关程度进行限定，仅在上面确定两个历史搜索项相关联时，还需要通过两个历史搜索项所属的领域来确定其关联程度。
在框606处，计算设备208基于多个历史搜索记录210之间的关联程度，确定多个历史搜索记录210之间的关系。计算设备208 基于所确定的关联程度，确定多个历史搜索记录210之间的关联关系。
在一些实施例中，多个历史搜索记录210及多个历史搜索记录210之间的关联关系可以由其他设备生成，计算设备208从其他设备获得。
在一些实施例中，不可以根据用户的搜索和浏览序列，提取重要关键词，建立用户关注点标签。
通过该方法，可以快速准确地确定出多个历史搜索记录之间的关联关系，使得可以快速准确的确定出扩展结果。
上面结合图6描述了根据本公开的一些实施例的用于确定历史搜索记录间的关系600的流程图。下面结合图7详细描述根据本公开的一些实施例的用于搜索内容的装置700的框图。
装置700包括优质结果筛选模块702、相关需求挖掘模块704、推荐结果匹配模块706和搜索结果与推荐结果混排模块708。在用户提交搜索词之后，相关需求挖掘模块704基于原始搜索词，挖掘出相关需求，相关需求的具体表达形式可以是搜索词的形式，也可以是关键词组合、语义向量等形式。基于挖掘出的相关需求，推荐结果匹配模块706检索优质结果筛选模块702筛选出的结果，从中找到可以满足相关需求的资源作为推荐结果；最后搜索结果与推荐结果混排模块708将推荐结果与搜索引擎检索出的正常结果进行混排，形成最终的结果列表，返回给用户。
相关需求挖掘模块704基于用户的原始搜索词挖掘出相关需求，其使用的技术方法包括如下几种：基于内容的挖掘方法：首先，对搜索项内容做拆分，定义了两种概念：搜索关键实体和需求维度。搜索核心主体是用户在搜索过程中，可以从搜索序列中提取的主体字串，这个主体字串能表征用户的核心诉求。如搜索项为“奔驰c200多少钱”，核心主体是“奔驰c200”，而“多少钱”是用户对核心主体的一种需求刻画，这里需求是询问价格。对于“奔驰c200”这个主体存在多种需求维度，如：奔驰c200图片，奔驰c200性能油 耗，奔驰c200销量等。基于内容的挖掘的思想是在保持关键实体不变的情况下，为用户推荐与搜索项本身维度强关联的不同需求维度的文章。
在挖掘核心主体时，首先从搜索日志中获取历史搜索项集合，通过命名实体识别(NER)和统计高频子串的方式确定关键实体，关键实体满足三个条件：1)关键实体本身作为搜索项有较高数量；2)关键实体作为子串频繁出现在多个历史搜索项中。3)在所有包含主体字串的搜索项中，主体字串的平均词项权重占比较高。
在挖掘需求维度时，需求维度是关键产体的属性，通过聚集同关键实体的搜索项，将去掉关键实体剩余的字串作为需求。初步获取的需求子串不同表述可能是同需求的，比如“奔驰c200多少钱”和“奔驰c200价格”的需求是相同的。我们通过聚类方法对需求子串做聚集，实现对不同需求类别的划分。同时根据不同类别的共现关联，计算不同维度之间的关联矩阵，以表征维度之间的紧密关系。
历史搜索项通过关键实体和需求信息的拆分，并通过挖掘好的需求类别关联矩阵，将强关联需求的搜索项作为当前搜索项的相关扩展需求集合。
还可以基于用户搜索大数据的挖掘方法来确定扩展历史搜索项：以所有用户搜索的搜索项为节点构建一张关联图，其中图的边包括：搜索行为(共现，具有相同检索结果的搜索项等)和领域关联(同领域，有强关联的不同领域)。基于图的挖掘方法，挖掘与当前搜索项强关联的搜索项集合，作为当前搜索项的相关扩展需求。同时根据用户的搜索和浏览序列，提取核心关键词，建立用户关注点标签。
推荐结果匹配模块706基于挖掘出的相关需求，从资源库中匹配可以满足相关需求的结果，其使用的技术方法包括如下几种：基于搜索检索系统的匹配：用扩展搜索项搜索检索系统，获取与扩展搜索项匹配的满足结果。并根据关联强度归并所有的结果，作为搜索项的推荐结果。基于用户搜索大数据的匹配：根据共现和有点等 用户行为，挖掘扩展搜索项关联的文章，作为搜索项的推荐结果。基于用户搜索和信息流浏览大数据的匹配：通过从用户搜索和信息流浏览数据中挖掘统计用户的关注点标签，并通过关注点匹配召回文章，作为用户的个性化推荐结果。
目标搜索结果与扩展结果混排模块708主要包括搜索结果打分、推荐结果打分、混排。搜索结果打分：主要基于历史点击分布、用户点击率预估等特征的融合模型打分。推荐结果打分基于历史点击分布、用户点击率预估打分等。
在混排时，基于搜索结果打分、推荐结果打分进行从高到低的排序。同时，也会进行多样性控制，包括基于推荐结果密度的多样性控制，以及同主题的推荐结果密度的多样性控制。
优质结果筛选模块702基于一些基本的质量因素，对文档资源进行打分，并基于打分进行截断，获取优质候选结果。基本质量因素包括：媒体站点打分：包括基于自动链接分析方法的站点打分，以及专家标注的站点打分；媒体作者打分：包括专家标注的作者登记、通过大数据分析的作者知名度、通过点赞、评论等读者反馈信息综合出的作者受欢迎度；媒体文本、图片、视频的丰富程度。
图8示出了根据本公开实施例的用于搜索内容的装置800的示意性框图。如图8所示，装置800可以包括历史搜索记录获取模块802，被配置为响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项。装置800还包括目标搜索项匹配模块804，被配置为从多个历史搜索记录中确定与目标搜索项匹配的第一历史搜索记录。装置800还包括历史搜索记录确定模块806，被配置为基于多个历史搜索记录之间的关系，从多个历史搜索记录中确定与第一历史搜索记录相关联的第二历史搜索记录。装置800还包括扩展结果确定模块，被配置为基于与第二历史搜索记录相对应的搜索结果，确定针对目标搜索项的扩展结果。
在一些实施例中，历史搜索记录获取模块802包括：第一历史 搜索项确定模块，被配置为从搜索日志中确定一组历史搜索请求所针对的一组历史搜索项；实体确定模块，被配置为从一组历史搜索项中确定多个实体，每个实体标识与对应历史搜索项相关联的对象；第一关键实体确定模块，被配置为基于多个实体在一组历史搜索项中的出现次数，从多个实体中确定关键实体；选择模块，被配置为从一组历史搜索项中选择包括关键实体的多个历史搜索项；以及生成模块，被配置为基于多个历史搜索项和关键实体生成多个历史搜索记录。
在一些实施例中，第一关键实体确定模块包括历史搜索项集合确定模块，被配置为从一组历史搜索项中确定包括单个实体的历史搜索项集合；第二历史搜索项确定模块，被配置为从历史搜索项集合确定至少一个历史搜索项，至少一个历史搜索项包括的单个实体在历史搜索项集合中的出现次数超过第一阈值次数；以及针对单个实体的关键实体确定模块，被配置为将至少一个历史搜索项包括的单个实体确定为关键实体。
在一些实施例中，关键实体确定模块包括高频实体确定模块，被配置为基于多个实体在一组历史搜索项中的出现次数，从多个实体中确定出现次数超过第二阈值次数的高频实体；以及第二关键实体确定模块，被配置为根据确定高频实体在对应的历史搜索项中的权重超过阈值权重，将高频实体确定为关键实体，其中权重指示高频实体在对应的历史搜索项中的重要性。
在一些实施例中，第二关键实体确定模块包括位置确定模块，被配置为高频实体在对应的历史搜索项中的位置，以及长度关系确定模块，被配置为高频实体的长度与对应的历史搜索项的长度之间的关系。
在一些实施例中，装置800还包括类别确定模块，被配置为基于多个历史搜索记录包括的多个历史搜索项，确定多个历史搜索记录的类别；以及历史搜索记录关系确定模块，被配置为基于类别，确定多个历史搜索记录之间的关系。
在一些实施例中，多个历史搜索记录中的每个历史搜索记录还包括关键实体，其中类别确定模块包括：剩余部分确定模块，被配置为通过从多个历史搜索项中去除相应的关键实体，获得多个历史搜索项各自的剩余部分；需求信息确定模块，被配置为至少基于剩余部分确定与多个历史搜索项相关联的需求信息；以及历史搜索记录类别确定模块，被配置为基于需求信息来确定多个历史搜索记录的类别。
在一些实施例中，多个历史搜索记录具有多个类别，并且其中历史搜索记录关系确定模块包括：第一搜索时间或搜索结果确定模块，被配置为从搜索日志中确定针对多个历史搜索项的搜索时间或搜索结果；相关程度确定模块，被配置为基于搜索时间或搜索结果，确定多个类别之间的相关程度；以及基于相关程度的关系确定模块，被配置为基于多个类别之间的相关程度，确定多个历史搜索记录之间的关系。
在一些实施例中，装置800还包括第二搜索时间或搜索结果配置模块，被配置为从搜索日志中确定针对多个历史搜索项的搜索时间或搜索结果；关联程度确定模块，被配置为基于搜索时间或搜索结果，确定多个历史搜索记录的之间的关联程度；以及基于关联程度的关系确定模块，被配置为基于多个历史搜索记录之间的关联程度，确定多个历史搜索记录之间的关系。
在一些实施例中，多个历史搜索记录中的每个历史搜索记录还包括关键实体和每个历史搜索记录的类别，其中历史搜索记录确定模块806包括：第二类别确定模块，被配置为基于多个历史搜索记录之间的关系，确定与第一历史搜索记录的第一类别相关联的第二类别；以及具有类别的第二历史搜索记录确定模块，被配置为从多个历史搜索记录确定具有第二类别的第二历史搜索记录，第二历史搜索记录包括第一历史搜索记录的关键实体。
在一些实施例中，历史搜索记录确定模块806包括一组历史搜索记录确定模块，被配置为基于多个历史搜索记录之间的关系，确 定与第一历史搜索记录相关联的一组历史搜索记录，第一历史搜索记录与一组历史搜索记录中的每个历史搜索记录具有关联程度；以及基于关联程度的历史搜索记录确定模块，被配置为基于关联程度，从一组历史搜索记录中确定第二历史搜索记录。
在一些实施例中，扩展结果确定模块808包括第一搜索结果获取模块，被配置为获取针对第二历史搜索记录中的历史搜索项的搜索结果；以及针对搜索结果的扩展结果确定模块，被配置为将搜索结果确定为扩展结果。
在一些实施例中，扩展结果确定模块808包括第二搜索结果获取模块，被配置为获取针对第二历史搜索记录中的历史搜索项的历史搜索结果；部分历史搜索结果确定模块，被配置为从历史搜索结果中确定已被用户访问的部分历史搜索结果；以及部分历史搜索结果扩展模块，被配置为将部分历史搜索结果确定为扩展结果。
在一些实施例中，扩展结果确定模块包括信息流模块，被配置为获取用户在针对第二历史搜索记录中的历史搜索项进行搜索时所产生的信息流；以及针地信息流的扩展结果模块，被配置为基于信息流，确定扩展结果。
在一些实施例中，装置800还包括以下至少一项：第一提供装置，被配置为提供扩展结果；以及第二提供装置，被配置为提供扩展结果和针对目标搜索项的目标搜索结果。
在一些实施例中，第二提供装置包括第一分数确定模块，被配置为确定扩展结果的第一分数，第一分数指示扩展结果与第二历史搜索记录中的历史搜索项的相关度；第二分数确定模块，被配置为确定目标搜索结果的第二分数，第二分数指示目标搜索结果与目标搜索项的相关度；优先级确定模块，被配置为基于第一分数和第二分数，确定扩展结果和目标搜索结果的优先级；以及扩展结果和搜索结果提供模块，被配置为基于优先级提供扩展结果和目标搜索结果。
在一些实施例中，装置800还包括目标数据源建立模块，被配 置为建立用于获得与第二历史搜索记录相对应的搜索结果的目标数据源。
在一些实施例中，目标数据源建立模块包括文档分数确定模块，被配置为确定多个原始数据源中的多个文档的分数，每个文档的分数指示文档的质量；以及目标数据源文档确定模块，被配置为将多个文档中分数超过阈值分数的文档确定为目标数据源中的文档。
图9示出了可以用来实施本公开的实施例的电子设备900的示意性框图。设备900可以用于实现图1中的终端设备204和计算设备208。如图所示，设备900包括计算单元901，其可以根据存储在只读存储器(ROM)902中的计算机程序指令或者从存储单元808加载到随机访问存储器(RAM)903中的计算机程序指令，来执行各种适当的动作和处理。在RAM 903中，还可存储设备900操作所需的各种程序和数据。计算单元901、ROM 902以及RAM 903通过总线904彼此相连。输入/输出(I/O)接口905也连接至总线904。
设备900中的多个部件连接至I/O接口905，包括：输入单元906，例如键盘、鼠标等；输出单元907，例如各种类型的显示器、扬声器等；存储单元908，例如磁盘、光盘等；以及通信单元909，例如网卡、调制解调器、无线通信收发机等。通信单元909允许设备900通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。
计算单元901可以是各种具有处理和计算能力的通用和/或专用处理组件。计算单元901的一些示例包括但不限于中央处理单元(CPU)、图形处理单元(GPU)、各种专用的人工智能(AI)计算芯片、各种运行机器学习模型算法的计算单元、数字信号处理器(DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元901执行上文所描述的各个方法和处理，例如方法300、400、500和600。例如，在一些实施例中，300、400、500和600可被实现为计算机软件程序，其被有形地包含于机器可读介质，例如存储单元908。在一些实施例中，计算机程序的部分或者全部可以经由ROM 902和/或通信单元909而被载入和/或安装到设备900上。当计算机程序加载到RAM 903并由计算单元901执行时，可以执行上文描述的方法300、400、500和600的一个或多个步骤。备选地，在其他实施例中，计算单元901可以通过其他任何适当的方式(例如，借助于固件)而被配置为执行方法900。
本文中以上描述的功能可以至少部分地由一个或多个硬件逻辑部件来执行。例如，非限制性地，可以使用的示范类型的硬件逻辑部件包括：场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、芯片上系统的系统(SOC)、负载可编程逻辑设备(CPLD)等等。
用于实施本公开的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器，使得程序代码当由处理器或控制器执行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行，作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或服务器上执行。
在本公开的上下文中，机器可读介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。
此外，虽然采用特定次序描绘了各操作，但是这应当理解为要 求这样操作以所示出的特定次序或以顺序次序执行，或者要求所有图示的操作应被执行以取得期望的结果。在一定环境下，多任务和并行处理可能是有利的。同样地，虽然在上面论述中包含了若干具体实现细节，但是这些不应当被解释为对本公开的范围的限制。在单独的实施例的上下文中描述的某些特征还可以组合地实现在单个实现中。相反地，在单个实现的上下文中描述的各种特征也可以单独地或以任何合适的子组合的方式实现在多个实现中。
尽管已经采用特定于结构特征和/或方法逻辑动作的语言描述了本主题，但是应当理解所附权利要求书中所限定的主题未必局限于上面描述的特定特征或动作。相反，上面所描述的特定特征和动作仅仅是实现权利要求书的示例形式。

Claims (38)

  1. 一种搜索内容的方法，包括：
    响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项；
    从所述多个历史搜索记录中确定与所述目标搜索项匹配的第一历史搜索记录；
    基于所述多个历史搜索记录之间的关系，从所述多个历史搜索记录中确定与所述第一历史搜索记录相关联的第二历史搜索记录；以及
    基于与所述第二历史搜索记录相对应的搜索结果，确定针对所述目标搜索项的扩展结果。
  2. 根据权利要求1所述的方法，还包括：
    基于所述多个历史搜索记录包括的多个历史搜索项，确定所述多个历史搜索记录的类别；以及
    基于所述类别，确定所述多个历史搜索记录之间的关系。
  3. 根据权利要求2所述的方法，其中所述多个历史搜索记录中的每个历史搜索记录还包括关键实体，其中确定所述多个历史搜索记录的类别包括：
    通过从所述多个历史搜索项中去除相应的关键实体，获得所述多个历史搜索项各自的剩余部分；
    至少基于所述剩余部分确定与所述多个历史搜索项相关联的需求信息；以及
    基于所述需求信息来确定所述多个历史搜索记录的类别。
  4. 根据权利要求2所述的方法，其中所述多个历史搜索记录具有多个类别，并且其中确定所述多个历史搜索记录之间的关系包括：
    从搜索日志中确定针对所述多个历史搜索项的搜索时间或搜索结果；
    基于所述搜索时间或所述搜索结果，确定所述多个类别之间的相关程度；以及
    基于所述多个类别之间的相关程度，确定所述多个历史搜索记录之间的关系。
  5. 根据权利要求1所述的方法，还包括：
    从搜索日志中确定针对所述多个历史搜索项的搜索时间或搜索结果；
    基于所述搜索时间或所述搜索结果，确定所述多个历史搜索记录的之间的关联程度；以及
    基于所述多个历史搜索记录之间的关联程度，确定所述多个历史搜索记录之间的关系。
  6. 根据权利要求1所述的方法，其中所述多个历史搜索记录中的每个历史搜索记录还包括关键实体和每个历史搜索记录的类别，其中确定所述第二历史搜索记录包括：
    基于所述多个历史搜索记录之间的所述关系，确定与所述第一历史搜索记录的第一类别相关联的第二类别；以及
    从所述多个历史搜索记录确定具有第二类别的第二历史搜索记录，所述第二历史搜索记录包括所述第一历史搜索记录的关键实体。
  7. 根据权利要求1所述的方法，其中确定所述第二历史搜索记录包括：
    基于所述多个历史搜索记录之间的关系，确定与所述第一历史搜索记录相关联的一组历史搜索记录，所述第一历史搜索记录与所述一组历史搜索记录中的每个历史搜索记录具有关联程度；以及
    基于所述关联程度，从所述一组历史搜索记录中确定所述第二历史搜索记录。
  8. 根据权利要求1所述的方法，其中获取所述多个历史搜索记录包括：
    从搜索日志中确定一组历史搜索请求所针对的一组历史搜索项；
    从所述一组历史搜索项中确定多个实体，每个实体标识与对应历史搜索项相关联的对象；
    基于所述多个实体在所述一组历史搜索项中的出现次数，从所述多个实体中确定关键实体；
    从所述一组历史搜索项中选择包括所述关键实体的多个历史搜索项；以及
    基于所述多个历史搜索项和所述关键实体生成所述多个历史搜索记录。
  9. 根据权利要求8所述的方法，其中确定所述关键实体包括：
    从所述一组历史搜索项中确定包括单个实体的历史搜索项集合；
    从所述历史搜索项集合确定至少一个历史搜索项，所述至少一个历史搜索项包括的单个实体在所述历史搜索项集合中的出现次数超过第一阈值次数；以及
    将所述至少一个历史搜索项包括的单个实体确定为所述关键实体。
  10. 根据权利要求8所述的方法，其中确定所述关键实体包括：
    基于所述多个实体在所述一组历史搜索项中的出现次数，从所述多个实体中确定出现次数超过第二阈值次数的高频实体；以及
    根据确定所述高频实体在对应的历史搜索项中的权重超过阈值权重，将所述高频实体确定为所述关键实体，其中所述权重指示所述高频实体在所述对应的历史搜索项中的重要性。
  11. 根据权利要求10所述的方法，其中所述权重是基于以下至少一项确定的：
    所述高频实体在所述对应的历史搜索项中的位置，以及
    所述高频实体的长度与所述对应的历史搜索项的长度之间的关系。
  12. 根据权利要求1所述的方法，其中确定针对所述目标搜索项的扩展结果包括：
    获取针对所述第二历史搜索记录中的历史搜索项的搜索结果；以及
    将所述搜索结果确定为所述扩展结果。
  13. 根据权利要求1所述的方法，其中确定针对所述目标搜索项的扩展结果包括：
    获取针对所述第二历史搜索记录中的历史搜索项的历史搜索结果；
    从所述历史搜索结果中确定已被用户访问的部分历史搜索结果；以及
    将所述部分历史搜索结果确定为所述扩展结果。
  14. 根据权利要求1所述的方法，其中确定针对所述目标搜索项的扩展结果包括：
    获取用户在针对所述第二历史搜索记录中的历史搜索项进行搜索时所产生的信息流；以及
    基于所述信息流，确定所述扩展结果。
  15. 根据权利要求1所述的方法，还包括以下至少一项：
    提供所述扩展结果；以及
    提供所述扩展结果和针对所述目标搜索项的目标搜索结果。
  16. 根据权利要求15所述的方法，其中提供所述扩展结果和所述目标搜索结果包括：
    确定所述扩展结果的第一分数，所述第一分数指示所述扩展结果与所述第二历史搜索记录中的历史搜索项的相关度；
    确定所述目标搜索结果的第二分数，所述第二分数指示所述目标搜索结果与所述目标搜索项的相关度；
    基于所述第一分数和所述第二分数，确定所述扩展结果和所述目标搜索结果的优先级；以及
    基于所述优先级提供所述扩展结果和所述目标搜索结果。
  17. 根据权利要求1所述的方法，还包括：
    建立用于获得与所述第二历史搜索记录相对应的搜索结果的目 标数据源。
  18. 根据权利要求17所述的方法，其中建立所述目标数据源包括：
    确定多个原始数据源中的多个文档的分数，每个文档的所述分数指示所述文档的质量；以及
    将所述多个文档中分数超过阈值分数的文档确定为所述目标数据源中的文档。
  19. 一种搜索内容的装置，包括：
    历史搜索记录获取模块，被配置为响应于接收到针对目标搜索项的搜索请求，获取与多个历史搜索请求有关的多个历史搜索记录，每个历史搜索记录包括相对应的历史搜索请求所针对的历史搜索项；
    目标搜索项匹配模块，被配置为从所述多个历史搜索记录中确定与所述目标搜索项匹配的第一历史搜索记录；
    历史搜索记录确定模块，被配置为基于所述多个历史搜索记录之间的关系，从所述多个历史搜索记录中确定与所述第一历史搜索记录相关联的第二历史搜索记录；以及
    扩展结果确定模块，被配置为基于与所述第二历史搜索记录相对应的搜索结果，确定针对所述目标搜索项的扩展结果。
  20. 根据权利要求19所述的装置，还包括：
    类别确定模块，被配置为基于所述多个历史搜索记录包括的多个历史搜索项，确定所述多个历史搜索记录的类别；以及
    历史搜索记录关系确定模块，被配置为基于所述类别，确定所述多个历史搜索记录之间的关系。
  21. 根据权利要求20所述的装置，其中所述多个历史搜索记录中的每个历史搜索记录还包括关键实体，其中所述类别确定模块包括：
    剩余部分确定模块，被配置为通过从所述多个历史搜索项中去除相应的关键实体，获得所述多个历史搜索项各自的剩余部分；
    需求信息确定模块，被配置为至少基于所述剩余部分确定与所述多个历史搜索项相关联的需求信息；以及
    历史搜索记录类别确定模块，被配置为基于所述需求信息来确定所述多个历史搜索记录的类别。
  22. 根据权利要求20所述的装置，其中所述多个历史搜索记录具有多个类别，并且其中所述历史搜索记录关系确定模块包括：
    第一搜索时间或搜索结果确定模块，被配置为从搜索日志中确定针对所述多个历史搜索项的搜索时间或搜索结果；
    相关程度确定模块，被配置为基于所述搜索时间或所述搜索结果，确定所述多个类别之间的相关程度；以及
    基于相关程度的关系确定模块，被配置为基于所述多个类别之间的相关程度，确定所述多个历史搜索记录之间的关系。
  23. 根据权利要求19所述的装置，还包括：
    第二搜索时间或搜索结果配置模块，被配置为从搜索日志中确定针对所述多个历史搜索项的搜索时间或搜索结果；
    关联程度确定模块，被配置为基于所述搜索时间或所述搜索结果，确定所述多个历史搜索记录的之间的关联程度；以及
    基于关联程度的关系确定模块，被配置为基于所述多个历史搜索记录之间的关联程度，确定所述多个历史搜索记录之间的关系。
  24. 根据权利要求19所述的装置，其中所述多个历史搜索记录中的每个历史搜索记录还包括关键实体和每个历史搜索记录的类别，其中所述历史搜索记录确定模块包括：
    第二类别确定模块，被配置为基于所述多个历史搜索记录之间的所述关系，确定与所述第一历史搜索记录的第一类别相关联的第二类别；以及
    具有类别的第二历史搜索记录确定模块，被配置为从所述多个历史搜索记录确定具有第二类别的第二历史搜索记录，所述第二历史搜索记录包括所述第一历史搜索记录的关键实体。
  25. 根据权利要求19所述的装置，其中所述历史搜索记录确定 模块包括：
    一组历史搜索记录确定模块，被配置为基于所述多个历史搜索记录之间的关系，确定与所述第一历史搜索记录相关联的一组历史搜索记录，所述第一历史搜索记录与所述一组历史搜索记录中的每个历史搜索记录具有关联程度；以及
    基于关联程度的历史搜索记录确定模块，被配置为基于所述关联程度，从所述一组历史搜索记录中确定所述第二历史搜索记录。
  26. 根据权利要求19所述的装置，其中所述历史搜索记录获取模块包括：
    第一历史搜索项确定模块，被配置为从搜索日志中确定一组历史搜索请求所针对的一组历史搜索项；
    实体确定模块，被配置为从所述一组历史搜索项中确定多个实体，每个实体标识与对应历史搜索项相关联的对象；
    第一关键实体确定模块，被配置为基于所述多个实体在所述一组历史搜索项中的出现次数，从所述多个实体中确定关键实体；
    选择模块，被配置为从所述一组历史搜索项中选择包括所述关键实体的多个历史搜索项；以及
    生成模块，被配置为基于所述多个历史搜索项和所述关键实体生成所述多个历史搜索记录。
  27. 根据权利要求26所述的装置，其中所述第一关键实体确定模块包括：
    历史搜索项集合确定模块，被配置为从所述一组历史搜索项中确定包括单个实体的历史搜索项集合；
    第二历史搜索项确定模块，被配置为从所述历史搜索项集合确定至少一个历史搜索项，所述至少一个历史搜索项包括的单个实体在所述历史搜索项集合中的出现次数超过第一阈值次数；以及
    针对单个实体的关键实体确定模块，被配置为将所述至少一个历史搜索项包括的单个实体确定为所述关键实体。
  28. 根据权利要求26所述的装置，其中关键实体确定模块包括：
    高频实体确定模块，被配置为基于所述多个实体在所述一组历史搜索项中的出现次数，从所述多个实体中确定出现次数超过第二阈值次数的高频实体；以及
    第二关键实体确定模块，被配置为根据确定所述高频实体在对应的历史搜索项中的权重超过阈值权重，将所述高频实体确定为所述关键实体，其中所述权重指示所述高频实体在所述对应的历史搜索项中的重要性。
  29. 根据权利要求28所述的装置，其中所述第二关键实体确定模块包括：
    位置确定模块，被配置为所述高频实体在所述对应的历史搜索项中的位置，以及
    长度关系确定模块，被配置为所述高频实体的长度与所述对应的历史搜索项的长度之间的关系。
  30. 根据权利要求19所述的装置，其中所述扩展结果确定模块包括：
    第一搜索结果获取模块，被配置为获取针对所述第二历史搜索记录中的历史搜索项的搜索结果；以及
    针对搜索结果的扩展结果确定模块，被配置为将所述搜索结果确定为所述扩展结果。
  31. 根据权利要求19所述的装置，其中所述扩展结果确定模块包括：
    第二搜索结果获取模块，被配置为获取针对所述第二历史搜索记录中的历史搜索项的历史搜索结果；
    部分历史搜索结果确定模块，被配置为从所述历史搜索结果中确定已被用户访问的部分历史搜索结果；以及
    部分历史搜索结果扩展模块，被配置为将所述部分历史搜索结果确定为所述扩展结果。
  32. 根据权利要求19所述的装置，其中所述扩展结果确定模块包括：
    信息流模块，被配置为获取用户在针对所述第二历史搜索记录中的历史搜索项进行搜索时所产生的信息流；以及
    针地信息流的扩展结果模块，被配置为基于所述信息流，确定所述扩展结果。
  33. 根据权利要求19所述的装置，还包括以下至少一项：
    第一提供装置，被配置为提供所述扩展结果；以及
    第二提供装置，被配置为提供所述扩展结果和针对所述目标搜索项的目标搜索结果。
  34. 根据权利要求33所述的装置，其中第二提供装置包括：
    第一分数确定模块，被配置为确定所述扩展结果的第一分数，所述第一分数指示所述扩展结果与所述第二历史搜索记录中的历史搜索项的相关度；
    第二分数确定模块，被配置为确定所述目标搜索结果的第二分数，所述第二分数指示所述目标搜索结果与所述目标搜索项的相关度；
    优先级确定模块，被配置为基于所述第一分数和所述第二分数，确定所述扩展结果和所述目标搜索结果的优先级；以及
    扩展结果和搜索结果提供模块，被配置为基于所述优先级提供所述扩展结果和所述目标搜索结果。
  35. 根据权利要求19所述的装置，还包括：
    目标数据源建立模块，被配置为建立用于获得与所述第二历史搜索记录相对应的搜索结果的目标数据源。
  36. 根据权利要求35所述的装置，其中所述目标数据源建立模块包括：
    文档分数确定模块，被配置为确定多个原始数据源中的多个文档的分数，每个文档的所述分数指示所述文档的质量；以及
    目标数据源文档确定模块，被配置为将所述多个文档中分数超过阈值分数的文档确定为所述目标数据源中的文档。
  37. 一种电子设备，包括：
    一个或多个处理器；以及
    存储装置，用于存储一个或多个程序，当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现根据权利要求1-18中任一项所述的方法。
  38. 一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序在被处理器执行时实现根据权利要求1-18中任一项所述的方法。
PCT/CN2020/117129 2020-04-01 2020-09-23 用于搜索内容的方法、装置、设备和计算机可读存储介质 WO2021196541A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP20929634.2A EP4113329A4 (en) 2020-04-01 2020-09-23 METHOD, APPARATUS AND DEVICE FOR SEARCHING CONTENT, AND COMPUTER-READABLE STORAGE MEDIUM
KR1020227027825A KR20220119745A (ko) 2020-04-01 2020-09-23 콘텐츠를 검색하는 방법, 장치, 기기 및 컴퓨터 판독 가능 저장 매체
JP2022553192A JP7451747B2 (ja) 2020-04-01 2020-09-23 コンテンツを検索する方法、装置、機器及びコンピュータ読み取り可能な記憶媒体
US17/914,557 US20230147941A1 (en) 2020-04-01 2020-09-23 Method, apparatus and device used to search for content

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010252907.2 2020-04-01
CN202010252907.2A CN111475725B (zh) 2020-04-01 2020-04-01 用于搜索内容的方法、装置、设备和计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2021196541A1 true WO2021196541A1 (zh) 2021-10-07

Family

ID=71749483

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/117129 WO2021196541A1 (zh) 2020-04-01 2020-09-23 用于搜索内容的方法、装置、设备和计算机可读存储介质

Country Status (6)

Country Link
US (1) US20230147941A1 (zh)
EP (1) EP4113329A4 (zh)
JP (1) JP7451747B2 (zh)
KR (1) KR20220119745A (zh)
CN (1) CN111475725B (zh)
WO (1) WO2021196541A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116628129A (zh) * 2023-07-21 2023-08-22 南京爱福路汽车科技有限公司 一种汽车配件搜索方法及系统

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111475725B (zh) * 2020-04-01 2023-11-07 百度在线网络技术（北京）有限公司 用于搜索内容的方法、装置、设备和计算机可读存储介质
CN112053688B (zh) * 2020-08-27 2024-03-08 海信视像科技股份有限公司 一种语音交互方法及交互设备、服务器
CN112528144A (zh) * 2020-12-08 2021-03-19 北京百度网讯科技有限公司 搜索推荐方法、装置、智能设备、电子设备及存储介质
CN113051485B (zh) * 2021-03-26 2023-08-22 北京达佳互联信息技术有限公司 群组搜索方法、装置、终端及存储介质
CN117972222B (zh) * 2024-04-02 2024-06-21 紫金诚征信有限公司 基于人工智能的企业信息检索方法及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577489A (zh) * 2012-08-08 2014-02-12 百度在线网络技术（北京）有限公司 一种网页浏览历史查询方法及装置
CN103617266A (zh) * 2013-12-03 2014-03-05 北京奇虎科技有限公司 个性化扩展搜索方法及装置、系统
CN106096003A (zh) * 2014-12-26 2016-11-09 奇飞翔艺（北京）软件有限公司 数据搜索方法及客户端
CN106372231A (zh) * 2016-09-08 2017-02-01 乐视控股（北京）有限公司 一种搜索方法及装置
CN111475725A (zh) * 2020-04-01 2020-07-31 百度在线网络技术（北京）有限公司 用于搜索内容的方法、装置、设备和计算机可读存储介质

Family Cites Families (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH01233517A (ja) * 1988-03-14 1989-09-19 Mitsubishi Electric Corp データベース検索装置
US7562069B1 (en) * 2004-07-01 2009-07-14 Aol Llc Query disambiguation
US20060224583A1 (en) * 2005-03-31 2006-10-05 Google, Inc. Systems and methods for analyzing a user's web history
CN101192223A (zh) * 2006-11-27 2008-06-04 北京三星通信技术研究有限公司 黄页搜索方法和黄页搜索系统
KR101348598B1 (ko) * 2007-12-21 2014-01-07 삼성전자주식회사 디지털 티비 방송 제공 시스템과 디지털 티비 및 그 제어방법
JP2010122932A (ja) * 2008-11-20 2010-06-03 Nippon Telegr & Teleph Corp <Ntt> 文書検索装置、文書検索方法、および文書検索プログラム
JP5220659B2 (ja) * 2009-02-27 2013-06-26 ヤフー株式会社 検索装置及び方法
US20100332493A1 (en) * 2009-06-25 2010-12-30 Yahoo! Inc. Semantic search extensions for web search engines
CN102012900B (zh) * 2009-09-04 2013-01-30 阿里巴巴集团控股有限公司 信息检索方法和系统
JP5493845B2 (ja) * 2009-12-28 2014-05-14 富士通株式会社 検索支援プログラム、検索支援装置、及び検索支援方法
CN101840420B (zh) * 2010-04-02 2011-12-28 清华大学 搜索辅助系统与搜索辅助方法
US20140358971A1 (en) * 2010-10-19 2014-12-04 Google Inc. Techniques for identifying chain businesses and queries
US20120203751A1 (en) * 2011-02-07 2012-08-09 International Business Machines Corporation Capture, Aggregate, and Use Search Activities as a Source of Social Data Within an Enterprise
KR101818717B1 (ko) * 2011-09-27 2018-01-15 네이버 주식회사 컨셉 키워드 확장 데이터 셋을 이용한 검색방법, 장치 및 컴퓨터로 판독 가능한 기록매체
AU2012247097B2 (en) * 2011-11-14 2015-04-16 Google Inc. Visual search history
CN102419776A (zh) * 2011-12-31 2012-04-18 北京百度网讯科技有限公司 一种满足用户多维度搜索需求的方法和设备
CN102929966B (zh) * 2012-10-12 2016-03-09 合一网络技术(北京)有限公司 一种用于提供个性化搜索列表的方法及系统
CN103049495A (zh) * 2012-12-07 2013-04-17 百度在线网络技术（北京）有限公司 用于提供与查询序列相对应的搜索建议的方法、装置与设备
CN103593410B (zh) * 2013-10-22 2017-04-12 上海交通大学 通过替换概念性词语进行搜索推荐的系统
US20160306887A1 (en) * 2013-12-03 2016-10-20 Beijing Qihoo Technology Company Limited Methods, apparatuses and systems for linked and personalized extended search
CN105335391B (zh) * 2014-07-09 2019-02-15 阿里巴巴集团控股有限公司 基于搜索引擎的搜索请求的处理方法和装置
CN104462325B (zh) * 2014-12-02 2019-05-03 百度在线网络技术（北京）有限公司 搜索推荐方法及装置
CN105893397B (zh) * 2015-06-30 2019-03-15 北京爱奇艺科技有限公司 一种视频推荐方法及装置
JP6664599B2 (ja) * 2015-08-25 2020-03-13 ヤフー株式会社 曖昧性評価装置、曖昧性評価方法、及び曖昧性評価プログラム
CN105898423A (zh) * 2015-12-08 2016-08-24 乐视网信息技术（北京）股份有限公司 视频推送方法、系统及服务器
CN106095819A (zh) * 2016-05-31 2016-11-09 北京奇艺世纪科技有限公司 一种视频推荐方法及装置
CN109446402B (zh) * 2017-08-29 2022-04-01 阿里巴巴集团控股有限公司 一种搜索方法及装置
CN108399232A (zh) * 2018-02-13 2018-08-14 北京奇虎科技有限公司 一种信息推送方法、装置及电子设备
CN109101658B (zh) * 2018-08-31 2022-05-10 优视科技新加坡有限公司 信息搜索方法、装置及设备/终端/服务器
EP3918486A4 (en) * 2019-02-01 2022-10-19 Ancestry.com Operations Inc. SEARCH AND RANK RECORDS ACROSS DIFFERENT DATABASES
CN110245357B (zh) * 2019-06-26 2023-05-02 北京百度网讯科技有限公司 主实体识别方法和装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577489A (zh) * 2012-08-08 2014-02-12 百度在线网络技术（北京）有限公司 一种网页浏览历史查询方法及装置
CN103617266A (zh) * 2013-12-03 2014-03-05 北京奇虎科技有限公司 个性化扩展搜索方法及装置、系统
CN106096003A (zh) * 2014-12-26 2016-11-09 奇飞翔艺（北京）软件有限公司 数据搜索方法及客户端
CN106372231A (zh) * 2016-09-08 2017-02-01 乐视控股（北京）有限公司 一种搜索方法及装置
CN111475725A (zh) * 2020-04-01 2020-07-31 百度在线网络技术（北京）有限公司 用于搜索内容的方法、装置、设备和计算机可读存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4113329A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116628129A (zh) * 2023-07-21 2023-08-22 南京爱福路汽车科技有限公司 一种汽车配件搜索方法及系统
CN116628129B (zh) * 2023-07-21 2024-02-27 南京爱福路汽车科技有限公司 一种汽车配件搜索方法及系统

Also Published As

Publication number Publication date
EP4113329A1 (en) 2023-01-04
EP4113329A4 (en) 2024-04-24
CN111475725B (zh) 2023-11-07
KR20220119745A (ko) 2022-08-30
US20230147941A1 (en) 2023-05-11
JP7451747B2 (ja) 2024-03-18
CN111475725A (zh) 2020-07-31
JP2023516209A (ja) 2023-04-18

Similar Documents

Publication Publication Date Title
WO2021196541A1 (zh) 用于搜索内容的方法、装置、设备和计算机可读存储介质
KR102075833B1 (ko) 미술 작품 추천 큐레이션 방법 및 시스템
US8095539B2 (en) Taxonomy-based object classification
US20190205472A1 (en) Ranking Entity Based Search Results Based on Implicit User Interactions
CN109271574A (zh) 一种热词推荐方法及装置
Xu et al. Web content mining
Hindle et al. Clustering web video search results based on integration of multiple features
US20150120720A1 (en) Method and system of identifying relevant content snippets that include additional information
US20100042610A1 (en) Rank documents based on popularity of key metadata
Makvana et al. A novel approach to personalize web search through user profiling and query reformulation
US20170185672A1 (en) Rank aggregation based on a markov model
Saoud et al. Integrating social profile to improve the source selection and the result merging process in distributed information retrieval
CN115248839A (zh) 一种基于知识体系的长文本检索方法以及装置
CN115391479A (zh) 用于文档搜索的排序方法、装置、电子介质及存储介质
Rajkumar et al. Users’ click and bookmark based personalization using modified agglomerative clustering for web search engine
Wang et al. An efficient refinement algorithm for multi-label image annotation with correlation model
Ren et al. Role-explicit query extraction and utilization for quantifying user intents
Rashmi et al. Deep web crawler: exploring and re-ranking of web forms
Ullah et al. Query subtopic mining for search result diversification
Kambau et al. Unified concept-based multimedia information retrieval technique
Hung et al. Reorganization of search results based on semantic clustering
Cheng et al. Learning To Rank Relevant Documents for Information Retrieval in Bioengineering Text Corpora
Sheth et al. Ontology Based Semantic Web Information Retrieval Enhancing Search Significance
Wang et al. GSR: A Resource Model and Semantics-based API Recommendation Algorithm
Nechakhin et al. Similar Papers Recommendation for Research Comparisons.

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: 20929634

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20227027825

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2022553192

Country of ref document: JP

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2020929634

Country of ref document: EP

Effective date: 20220930

NENP Non-entry into the national phase

Ref country code: DE