WO2017088496A1 - 一种搜索推荐方法、装置、设备及计算机存储介质 - Google Patents

一种搜索推荐方法、装置、设备及计算机存储介质 Download PDF

Info

Publication number
WO2017088496A1
WO2017088496A1 PCT/CN2016/089483 CN2016089483W WO2017088496A1 WO 2017088496 A1 WO2017088496 A1 WO 2017088496A1 CN 2016089483 W CN2016089483 W CN 2016089483W WO 2017088496 A1 WO2017088496 A1 WO 2017088496A1
Authority
WO
WIPO (PCT)
Prior art keywords
search
search result
recommendation information
recommendation
entity
Prior art date
Application number
PCT/CN2016/089483
Other languages
English (en)
French (fr)
Inventor
黄际洲
周里成
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
Publication of WO2017088496A1 publication Critical patent/WO2017088496A1/zh

Links

Images

Classifications

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

Definitions

  • the present invention relates to the field of Internet search technologies, and in particular, to a search recommendation method, apparatus, device, and computer storage medium.
  • the user after the user inputs a query word (Query) and clicks a certain search result in the search result page, and completes the related operation on the page of the search result, the user returns the query word from the page of the search result. Match the search results page and further continue on the search results page to complete the search task. After returning to the search result page, the user usually recommends relevant search suggestions below the clicked search result to reduce the search cost and improve the search efficiency.
  • Query query word
  • the user After returning to the search result page, the user usually recommends relevant search suggestions below the clicked search result to reduce the search cost and improve the search efficiency.
  • the embodiment of the present invention provides a search recommendation method and device, which improves the diversity of recommendation information in the search result page and the efficiency of obtaining information.
  • An aspect of an embodiment of the present invention provides a search recommendation method, including:
  • search result is opened in the search results page, and returned by the page of the search result Returning to the search result page to obtain at least two sets of recommendation information;
  • At least two of the description tags and at least two sets of the recommendation information are correspondingly displayed in the search result page.
  • the foregoing aspect and any possible implementation manner further provide an implementation manner, where the obtaining at least two sets of recommendation information includes:
  • a corresponding set of recommendation information is obtained according to each of the target search behaviors.
  • the aspect as described above, and any possible implementation manner, further provide an implementation manner, where the obtaining, according to the entity matched by the search result, obtaining at least two target search behaviors of the user, including:
  • a first recommendation model Entering, by the entity matching the search result, a first recommendation model, such that the first recommendation model obtains at least two candidate search behaviors that match the entity according to the entity, as the at least two Target search behavior and output;
  • the first recommendation model includes a correspondence between an entity and a candidate search behavior.
  • the aspect as described above, and any possible implementation manner further provide an implementation manner, according to each of the target search behaviors, obtaining a corresponding set of recommendation information, including:
  • the second recommendation model includes a correspondence between the candidate search behavior and the recommendation information.
  • At least two description tags are displayed, and a corresponding set of recommendation information is displayed in a designated area of each of the description tags.
  • An aspect of an embodiment of the present invention provides a search recommendation apparatus, including:
  • An information obtaining module configured to obtain at least two sets of recommendation information if a search result is opened in a search result page, and the search result page is returned from the page of the search result;
  • a label obtaining module configured to generate a corresponding description label for each group of the recommendation information
  • a recommendation display module configured to display at least two of the description tags and at least two sets of the recommendation information in the search result page.
  • a corresponding set of recommendation information is obtained according to each of the target search behaviors.
  • the information acquiring module is configured to obtain at least two target search behaviors of the user according to the entity matched by the search result, specifically to:
  • a first recommendation model Entering, by the entity matching the search result, a first recommendation model, such that the first recommendation model obtains at least two candidate search behaviors that match the entity according to the entity, as the at least two Target search behavior and output;
  • the first recommendation model includes a correspondence between an entity and a candidate search behavior.
  • the information obtaining module is configured to: when obtaining a corresponding set of recommendation information according to each of the target search behaviors, specifically:
  • the second recommendation model includes a correspondence between the candidate search behavior and the recommendation information.
  • recommendation display module is specifically configured to:
  • At least two description tags are displayed, and a corresponding set of recommendation information is displayed in a designated area of each of the description tags.
  • the technical solution provided by the embodiment of the present invention after the user clicks on a certain search result, multiple sets of recommendation information are provided to the user on the search result page, and the recommendation information group is displayed, and a description label is added for each group of recommendation information, so as to facilitate the user. According to the grouping of the description label and the recommendation information, the required recommendation information can be obtained in time, and the technical solution provided by the embodiment of the present invention is improved compared with the recommendation information that the user manually screens the recommendation information and provides a single dimension in the prior art. The diversity of recommended information in the search results page and the efficiency of obtaining information.
  • FIG. 1 is a schematic flowchart of a search recommendation method according to an embodiment of the present invention
  • FIG. 3 is a first exemplary diagram of recommended information displayed by an embodiment of the present invention.
  • FIG. 5 is a third exemplary diagram of recommended information displayed by an embodiment of the present invention.
  • FIG. 7 is a fourth exemplary diagram of recommended information displayed by an embodiment of the present invention.
  • FIG. 8 is a functional block diagram of a search recommendation apparatus according to an embodiment of the present invention.
  • first, second, etc. may be used to describe the recommendation models in embodiments of the invention, these recommendation models should not be limited to these terms. These terms are only used to distinguish recommendation models from each other.
  • the first recommendation model may also be referred to as a second recommendation model without departing from the scope of the embodiments of the present invention.
  • the second recommendation model may also be referred to as a first recommendation model.
  • the word “if” as used herein may be interpreted as “when” or “when” or “in response to determining” or “in response to detecting.”
  • the phrase “if determined” or “if detected (conditions or events stated)” may be interpreted as “when determined” or “in response to determination” or “when detected (stated condition or event) “Time” or “in response to a test (condition or event stated)”.
  • FIG. 1 it is a schematic flowchart of a search recommendation method according to an embodiment of the present invention. As shown in the figure, the method includes the following steps:
  • the query word input by the user may be received, and then searched according to the query word to obtain a plurality of search results that match the query word, and finally, the search result page is used to display a plurality of search results to the user.
  • the user clicks on the Uniform Resource Locator (URL) of a search result in the search result page the page of the search result is opened in the search result page, so that the user can browse the page of the search result, and then, After the user browses the page of the search result, the page of the search result returns to the previous search result page.
  • URL Uniform Resource Locator
  • the method of obtaining at least two sets of recommendation information may include, but is not limited to:
  • At least two target search behaviors of the user are obtained according to the entity matched by the search result. Then, according to each of the target search behaviors, a corresponding set of recommendation information is obtained.
  • the target search behavior is specified to be a possible subsequent search behavior of the user.
  • the search behavior obtained after the line prediction analysis belongs to the search behavior that the user wants to complete the search task.
  • the method for obtaining at least two target search behaviors of the user according to the entity matched by the search result may include, but is not limited to:
  • a first recommendation model is generated, the first recommendation model including a correspondence between an entity and a candidate search behavior. Then, an entity matching the search result page is input into the first recommendation model such that the first recommendation model obtains at least two candidate search behaviors matching the entity according to the entity, as the At least two target search behaviors and output.
  • the first recommendation model may directly use the query word as an entity, or may extract an entity from the query word; the first recommendation model may be based on the obtained entity, in the entity and the candidate search behavior. Performing a search in the correspondence relationship, so that at least two candidate search behaviors matching the entity may be obtained, the two candidate search behaviors may be at least two target search behaviors of the user, and the first recommendation entity outputs at least two target search behaviors .
  • the method of generating the first recommendation model may include, but is not limited to:
  • the query word related to the search result clicked by the user in the search result page is obtained; then, the entity in the obtained query word is extracted, and the extracted entity is connected to the corresponding other entity, and finally extracted according to The entity and other entities perform search behavior analysis, and generate correspondence between the entity and the candidate search behavior according to the analysis result.
  • a user clicks on a search result in a search result page that matches a query term, and the query term can then serve as a query term associated with the search result.
  • the user inputs another query word, and the other query words can be used as query terms related to the search result clicked by the user.
  • the two methods can be used to continuously search for query words related to the search result, and correspondingly store the search result and the related query words, so that the matching query can be performed in the corresponding relationship according to the search result clicked by the user, Get the query terms related to the search results that users clicked on the search results page.
  • the correspondence between the entity and the candidate search behavior may be ⁇ Qian Mapo, then search for Hayao Miyazaki, 0.85> and ⁇ Min's mother, then search for Spirited Away, 0.59>, where "Qian Po Po" is an entity, "Next search for person name A” and “Next search for anime B” are candidate search behaviors, and the value in the correspondence relationship is a probability value, which means that from the click entity "Qian Po Po" to the next "Next search for Hayao Miyazaki" The probability of this candidate search behavior, and the probability of the candidate search behavior from the click entity "Qian Po Po" to the next "Search for Spirited Away", the higher the probability value, the more likely the user is after clicking on the entity. Perform subsequent search behavior.
  • the user after the user searches for a query word, clicks on a search result in the search result page, and returns to the original search result page, the user may analyze and predict the subsequent possible search behavior. Users want to complete the search behavior that their search behavior may perform, and then search and recommend based on the predicted search behavior.
  • a corresponding set of recommendation information may be obtained according to each target search behavior, and a set of recommendation information may include a certain number of recommendation information.
  • the method for obtaining a corresponding set of recommendation information according to each target search behavior may include, but is not limited to:
  • Each of the target search behaviors may be input to a second recommendation model, the second recommendation model including a correspondence between the candidate search behavior and the recommendation information, the second recommendation model being The target search behavior is searched, and the corresponding search information is obtained in the corresponding relationship between the candidate search behavior and the recommendation information, and the corresponding set of recommendation information is output.
  • a query word related to the search result is obtained in S101, and the query word can be used as recommendation information, and some query words related to the same search result can be used as recommendation information of the search result, and further A correspondence between the candidate search behavior and the recommendation information may be generated as the second recommendation model.
  • search for characters in thousands of thousands of plays a series of characters in the play, 0.8>, where "search for characters in thousands of plays" is a candidate search behavior, "a series of role names in the play” includes recommendation information 0.8 is a probability value indicating the probability of clicking the recommendation information after performing the candidate search behavior.
  • the third recommendation model may be generated according to the first recommendation model and the second recommendation model, and the entity matched by the clicked search result is input into the third recommendation model, and the third recommendation model is searched according to the entity to obtain And output a set of recommendation information corresponding to the entity.
  • the second recommendation model since the first recommendation model includes a correspondence between the entity and the candidate search behavior, the second recommendation model includes a correspondence between the candidate search behavior and the recommendation information, and therefore, the entity can be directly generated based on the foregoing two correspondences. Correspondence with recommendation information as a third recommendation model.
  • the correspondence between the entity included in the third recommendation model and the recommendation information may be:
  • recommendation information A recommendation information B, recommendation information C;
  • recommendation information H Look for related recommendations for Japanese anime: recommendation information H, recommendation information I, recommendation information J.
  • a method for generating a corresponding description label for each group of recommendation information may include, but is not limited to:
  • the first type First, determine the description keyword according to each target search behavior. Then, using the description keyword and the preset first label template, a corresponding description label is generated for each group of the recommendation information.
  • the target search behavior includes ⁇ Qian Po, then search for Spirited Away, 0.59>, and the description keyword is “Spirited Away”.
  • the default first label template can be “About xxx, everyone is searching. ", where xxx can be replaced with the determined description keyword, so that the description tag can be generated according to the description keyword and the first tag template, such as "About Spirit and Chihiro, everyone is searching”.
  • a set of recommendation information includes "Miyazaki Hayao”, “faceless man”, “Tang Po”, and the entity corresponding to the recommended information is "Spirited Away", and the preset second label template may be "xxx related", where xxx can be replaced with the determined entity, so that a description tag can be generated according to each set of recommendation information and the second tag template, such as "Thousands of Thousands Related”.
  • the two groups of recommendation information and the corresponding description label need to be correspondingly displayed in the search result page, so as to implement search recommendation to the user.
  • the method for correspondingly displaying at least two description labels and at least two sets of recommendation information in the search result page may include, but is not limited to:
  • At least two description tags are displayed, and a corresponding set of recommendation information is displayed in a designated area of each of the description tags.
  • the target search behavior includes three dimensions of A, B, and C, such as the attribute of the query word, the classification of the query word, and the upper layer requirement of the query word. These are the recommendation information that the user wants to obtain next.
  • the description tags are added to the target search behaviors of the three dimensions, and the description tags and the corresponding recommendation data groups are displayed to the user, so that the user can quickly locate the search recommendations of interest. In order to facilitate users to complete related search tasks, reduce search costs and improve search efficiency.
  • FIG. 2 is a first example diagram of the recommendation information displayed in the prior art.
  • the user inputs the query word “Qian Po Po” (Miyazaki’s Japanese anime work “Spirited Away”
  • the search results page matching "Qian Po Po” clicked the first search result, that is, "Qian Po Po” on the Baidu Encyclopedia page, where users can learn about Qian Popo Basic information about this role.
  • the page of Baidu Encyclopedia returns to the original search result page, as shown in FIG.
  • the recommended information displayed to the user in the prior art includes: the seal of Qian Po Po, Qian Po Po and Tang Po, Qian Qian Qian Qian Po, money Mother-in-law
  • the number of recommendations is relatively small, and the category of recommendation information is single.
  • FIG. 3 is a first example diagram of recommended information displayed according to an embodiment of the present invention.
  • the query word “Qian Po Po” is combined with the user click.
  • the user's possible follow-up search behavior is analyzed, and the user's next three possible search behaviors may be: understanding other information related to Qian's mother-in-law, understanding "thousands Find other characters in the anime and learn about similar Japanese anime.
  • each group of dimension information can include 4 to 6 recommendation information.
  • FIG. 4 is a second exemplary diagram of the recommended information displayed in the embodiment of the present invention.
  • the description tags "Meng Po Po Related”, “Thousands of Thousand Related” and “Japanese Anime Related” as shown in FIG. 4 can be employed.
  • FIG. 5 is a third embodiment of recommended information displayed according to an embodiment of the present invention.
  • the technical solution provided by the embodiment of the present invention is applied to a mobile phone, since the screen of the mobile phone is small, three description tags and one of the description tags corresponding to FIG. 5 can be displayed.
  • a set of recommendation information when the user clicks on a description tag, and then displays a group of recommendation information corresponding to the description tag, so that for each description tag, a greater number of recommendation information can be displayed.
  • FIG. 6 is a second example diagram of the recommendation information displayed in the prior art.
  • the search result matches the “Beijing Housing Provident Fund”.
  • the page of the official website of the Beijing Housing Provident Fund returns to the original search result page.
  • the recommended information displayed to the user includes: Beijing Housing Provident Fund inquiry password, 12329 ID card for the public accumulation fund, and two recommended information. Less, and the category of recommended information is single.
  • FIG. 7 is a fourth example diagram of recommended information displayed according to an embodiment of the present invention.
  • the query term “Beijing Housing Provident Fund” is combined with the user. After clicking on the search results of the Beijing Housing Provident Fund official website, the user's possible follow-up search behavior is analyzed.
  • the following three possible search behaviors that may be of interest to the user may include: 1) querying the amount of the reserve, but encountering the process of querying the provident fund The problem (such as forgetting the password), return to the original search results page to continue to complete the next step related to the provident fund inquiry; 2) use the provident fund to buy a house, and understand the issues related to the purchase of housing, 3) understand more about the purchase of housing related issues. Therefore, the recommendation information corresponding to the three search behaviors is ordered and grouped, and the description label "About the website landing problem, everyone is searching", "About the CPF purchase problem, everyone is searching" and “About the purchase problem" everyone is searching, and corresponding recommendation information, each group of dimension information can include 2 to 4 recommendation information.
  • Second-hand housing "reliable purchase agent” and "house with a total price of 2 million.”
  • the technical solution provided by the embodiment of the present invention not only provides the user with more recommendation information that may be of interest, but also enables the user to quickly locate the search recommendation of interest, so that the user can better complete the related search task and reduce the search. Cost, improve search efficiency.
  • Embodiments of the present invention further provide an apparatus embodiment for implementing the steps and methods in the foregoing method embodiments.
  • FIG. 8 is a functional block diagram of a search recommendation apparatus according to an embodiment of the present invention. As shown, the device includes:
  • the information obtaining module 81 is configured to obtain at least two sets of recommendation information if a search result is opened in the search result page, and the search result page is returned from the page of the search result;
  • the label obtaining module 82 is configured to generate a corresponding description label for each group of the recommendation information.
  • the recommendation display module 83 is configured to display at least two of the description tags and at least two sets of the recommendation information in the search result page.
  • the information obtaining module 81 is specifically configured to:
  • a corresponding set of recommendation information is obtained according to each of the target search behaviors.
  • the information obtaining module 81 is configured to perform the search according to the The result of the matching entity, when obtaining at least two target search behaviors of the user, specifically for:
  • a first recommendation model Entering, by the entity matching the search result, a first recommendation model, such that the first recommendation model obtains at least two candidate search behaviors that match the entity according to the entity, as the at least two Target search behavior and output;
  • the first recommendation model includes a correspondence between an entity and a candidate search behavior.
  • the information obtaining module 81 is configured to: when obtaining a corresponding set of recommendation information according to each of the target search behaviors, specifically:
  • the second recommendation model includes a correspondence between the candidate search behavior and the recommendation information.
  • the label obtaining module 82 is specifically configured to:
  • the label obtaining module 82 is specifically configured to:
  • the recommendation display module 83 is specifically configured to:
  • At least two description tags are displayed, and a corresponding set of recommendation information is displayed in a designated area of each of the description tags.
  • each unit in this embodiment can perform the method shown in FIG. 1, the embodiment is not detailed. For a detailed description, reference may be made to the related description of FIG.
  • a search result is opened in the search result page, and the search result page is returned from the page of the search result, at least two sets of recommendation information are obtained; thus, corresponding information is generated for each group of the recommendation information.
  • a description tag further, at least two of the description tags and at least two sets of the recommendation information are correspondingly displayed in the search result page.
  • the technical solution provided by the embodiment of the present invention after the user clicks on a certain search result, multiple sets of recommendation information are provided to the user on the search result page, and the recommendation information group is displayed, and a description label is added for each group of recommendation information, so as to facilitate the user. According to the grouping of the description label and the recommendation information, the required recommendation information can be obtained in time, and the technical solution provided by the embodiment of the present invention is improved compared with the recommendation information that the user manually screens the recommendation information and provides a single dimension in the prior art. The diversity of recommended information in the search results page and the efficiency of obtaining information.
  • the disclosed system, apparatus, and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • multiple units or components may be combined. Or it can be integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional units.
  • the above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium.
  • the above software functional unit is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. .

Abstract

一种搜索推荐方法、装置、设备及计算机存储介质。所述方法通过若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息(S101);从而,为每组所述推荐信息生成对应的描述标签(S102);进而,在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示(S103)。该方法提高了搜索结果页中推荐信息的多样性以及获取信息的效率。

Description

一种搜索推荐方法、装置、设备及计算机存储介质
本申请要求了申请日为2015年11月25日,申请号为201510828621.3发明名称为“一种搜索推荐方法及装置”的中国专利申请的优先权。
技术领域
本发明涉及互联网搜索技术领域,尤其涉及一种搜索推荐方法、装置、设备和计算机存储介质。
背景技术
现有技术中,用户在输入查询词(Query)并在搜索结果页中点击某条搜索结果之后,并在该条搜索结果的页面完成相关操作后,再从该搜索结果的页面返回与查询词相匹配的搜索结果页,并进一步会在搜索结果页继续进行相关操作,以完成搜索任务。其中,在返回该搜索结果页之后,目前通常会在点击的搜索结果的下方为用户推荐相关的搜索建议,以减少搜索成本,提高搜索效率。
然而,目前只能推荐较为热门的信息,往往是单一维度的推荐信息,而且需要用户在推荐信息中进行人工逐个进行查看,然后筛选出用户自身所需要的信息,因此现有技术中搜索结果页中的推荐信息的类型单一且获取信息的效率比较低。
发明内容
有鉴于此,本发明实施例提供了一种搜索推荐方法及装置,提高了搜索结果页中推荐信息的多样性以及获取信息的效率。
本发明实施例的一方面,提供一种搜索推荐方法,包括:
若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返 回所述搜索结果页,获得至少两组推荐信息;
为每组所述推荐信息生成对应的描述标签;
在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述获得至少两组推荐信息,包括:
根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为;
根据每个所述目标搜索行为,获得对应的一组推荐信息。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为,包括:
将与所述搜索结果相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出;
其中,所述第一推荐模型包括实体与候选搜索行为的对应关系。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,根据每个所述目标搜索行为,获得对应的一组推荐信息,包括:
将每个所述目标搜索行为输入第二推荐模型,以使得所述第二推荐模型根据输入的目标搜索行为,获得并输出对应的一组推荐信息;
其中,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,为每组所述推荐信息生成对应的描述标签,包括:
根据每个所述目标搜索行为,确定描述关键词;
利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,为每组所述推荐信息生成对应的描述标签,包括:
获得每组所述推荐信息对应的实体;
利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示,包括:
在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
本发明实施例的一方面,提供一种搜索推荐装置,包括:
信息获取模块,用于若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;
标签获取模块,用于为每组所述推荐信息生成对应的描述标签;
推荐显示模块,用于在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述信息获取模块,具体用于:
根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行 为;
根据每个所述目标搜索行为,获得对应的一组推荐信息。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述信息获取模块用于根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为时,具体用于:
将与所述搜索结果相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出;
其中,所述第一推荐模型包括实体与候选搜索行为的对应关系。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述信息获取模块用于根据每个所述目标搜索行为,获得对应的一组推荐信息时,具体用于:
将每个所述目标搜索行为输入第二推荐模型,以使得所述第二推荐模型根据输入的目标搜索行为,获得并输出对应的一组推荐信息;
其中,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述标签获取模块,具体用于:
根据每个所述目标搜索行为,确定描述关键词;
利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述标签获取模块,具体用于:
获得每组所述推荐信息对应的实体;
利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述推荐显示模块,具体用于:
在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
由以上技术方案可以看出,本发明实施例具有以下有益效果:
根据本发明实施例提供的技术方案,可以在用户点击某搜索结果后,在搜索结果页面上向用户提供多组推荐信息且推荐信息分组显示,并为每组推荐信息添加描述标签,以便于用户根据描述标签和推荐信息的分组,能够及时获得所需要的推荐信息,与现有技术中用户人工筛选推荐信息和提供单一维度的推荐信息相比,本发明实施例所提供的技术方案,提高了搜索结果页中推荐信息的多样性以及获取信息的效率。
附图说明
为了更清楚地说明本发明实施例的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其它的附图。
图1是本发明实施例所提供的搜索推荐方法的流程示意图;
图2是现有技术中显示的推荐信息的第一示例图;
图3是本发明实施例所提供的显示的推荐信息的第一示例图;
图4是本发明实施例所提供的显示的推荐信息的第二示例图;
图5是本发明实施例所提供的显示的推荐信息的第三示例图;
图6是现有技术中显示的推荐信息的第二示例图;
图7是本发明实施例所提供的显示的推荐信息的第四示例图;
图8是本发明实施例所提供的搜索推荐装置的功能方块图。
具体实施方式
为了使本发明的目的、技术方案和优点更加清楚,下面结合附图和具体实施例对本发明进行详细描述。
应当明确,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其它实施例,都属于本发明保护的范围。
在本发明实施例中使用的术语是仅仅出于描述特定实施例的目的,而非旨在限制本发明。在本发明实施例和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式,除非上下文清楚地表示其他含义。
应当理解,本文中使用的术语“和/或”仅仅是一种描述关联对象的关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中字符“/”,一般表示前后关联对象是一种“或”的关系。
应当理解,尽管在本发明实施例中可能采用术语第一、第二等来描述推荐模型,但这些推荐模型不应限于这些术语。这些术语仅用来将推荐模型彼此区分开。例如,在不脱离本发明实施例范围的情况下,第一推荐模型也可以被称为第二推荐模型,类似地,第二推荐模型也可以被称为第一推荐模型。
取决于语境,如在此所使用的词语“如果”可以被解释成为“在……时”或“当……时”或“响应于确定”或“响应于检测”。类似地,取决于语境,短语“如果确定”或“如果检测(陈述的条件或事件)”可以被解释成为“当确定时”或“响应于确定”或“当检测(陈述的条件或事件)时”或“响应于检测(陈述的条件或事件)”。
实施例一
本发明实施例给出一种搜索推荐方法,请参考图1,其为本发明实施例所提供的搜索推荐方法的流程示意图,如图所示,该方法包括以下步骤:
S101,若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息。
具体的,本发明实施例中,可以接收用户输入的查询词,然后根据该查询词进行搜索,以获得与查询词相匹配的若干搜索结果,最后通过搜索结果页向用户展现若干搜索结果。若用户在搜索结果页中点击某搜索结果的统一资源定位符(Uniform Resource Locator,URL),则在搜索结果页中打开该搜索结果的页面,这样,用户可以浏览该搜索结果的页面,接着,用户浏览完搜索结果的页面后,由搜索结果的页面返回之前的搜索结果页,此时,需要获得至少两组推荐信息。
举例说明,获得至少两组推荐信息的方法可以包括但不限于:
首先,根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为。然后,根据每个所述目标搜索行为,获得对应的一组推荐信息。
可以理解的是,目标搜索行为指定是对用户后续可能的搜索行为进 行预测分析后获得的搜索行为,属于用户想要完成搜索任务还需要进行的搜索行为。
举例说明,本发明实施例中,根据搜索结果所匹配的实体,获得用户的至少两个目标搜索行为的方法可以包括但不限于:
首先,生成第一推荐模型,第一推荐模型包括实体与候选搜索行为的对应关系。然后,将与所述搜索结果页相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出。
在一个具体的实现过程中,第一推荐模型可以直接将所述查询词作为实体,或者,也可以从查询词中提取实体;第一推荐模型可以根据获得的实体,在实体与候选搜索行为的对应关系中进行查找,从而可以获得与该实体相匹配的至少两个候选搜索行为,这两个候选搜索行为可以作为用户的至少两个目标搜索行为,第一推荐实体输出至少两个目标搜索行为。
举例说明,生成第一推荐模型的方法可以包括但不限于:
首先,获取与用户在搜索结果页中点击的搜索结果相关的查询词;然后,从获取的查询词中抽取出其中的实体,并将抽取出的实体连接到对应的其他实体,最后根据抽取出的实体以及其他实体,进行搜索行为分析,根据分析结果生成实体与候选搜索行为的对应关系。
例如,用户在与查询词相匹配的搜索结果页中点击了某个搜索结果,然后该查询词可以作为与该搜索结果相关的查询词。或者,又例如,用户点击了某个搜索结果后,又返回搜索结果页后,用户又输入了其他查询词,该其他查询词可以作为与用户点击的搜索结果相关的查询词。如 此,可以利用这两种方式来不断挖掘与搜索结果相关的查询词,进而对应存储搜索结果与相关的查询词,这样,就可以根据用户点击的搜索结果在该对应关系中进行匹配查询,以获得与用户在搜索结果页中点击的搜索结果相关的查询词。
例如,实体与候选搜索行为的对应关系可以为<钱婆婆,接下来搜索宫崎骏,0.85>和<钱婆婆,接下来搜索千与千寻,0.59>,其中,“钱婆婆”为实体,“接下来搜索人名A”和“接下来搜索动漫B”为候选搜索行为,对应关系中的数值为概率值,表示从点击实体“钱婆婆”,到接下来做“接下来搜索宫崎骏”这个候选搜索行为的概率,以及从点击实体“钱婆婆”,到接下来做“接下来搜索千与千寻”这个候选搜索行为的概率,概率值越高,表示用户在点击实体后越有可能执行后续的搜索行为。
可以理解的是,本发明实施例中,在用户搜索某查询词,并点击了搜索结果页中某个搜索结果,以及返回原搜索结果页后,可以对用户后续可能的搜索行为进行分析,预测用户想要完成自己的搜索行为可能会进行的搜索行为,进而基于预测的搜索行为进行搜索推荐。
本发明实施例中,在获得用户的至少两个目标搜索行为后,可以根据其中的每个目标搜索行为,都获得对应的一组推荐信息,一组推荐信息中可以包括若干数目的推荐信息。
举例说明,本发明实施例中,根据每个目标搜索行为,获得对应的一组推荐信息的方法可以包括但不限于:
可以将每个所述目标搜索行为输入第二推荐模型,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系,所述第二推荐模型根据输 入的目标搜索行为,在候选搜索行为与推荐信息的对应关系中进行查找,获得对应的一组推荐信息,并输出该对应的一组推荐信息。
在一个具体的实现过程中,S101中获得了与搜索结果相关的查询词,该查询词可以作为推荐信息,与同一个搜索结果相关的若干查询词,都可以作为该搜索结果的推荐信息,进而可以生成候选搜索行为与推荐信息的对应关系,以作为第二推荐模型。
例如,<搜索千与千寻剧中人物,一系列剧中角色名称,0.8>,其中,“搜索千与千寻剧中人物”为候选搜索行为,“一系列剧中角色名称”包括推荐信息,0.8为概率值,表示在进行该候选搜索行为后,点击推荐信息的概率。
或者,也可以根据第一推荐模型和第二推荐模型,生成第三推荐模型,并将点击的搜索结果所匹配的实体,输入第三推荐模型,第三推荐模型根据该实体进行查找,以获得并输出与该实体对应的一组推荐信息。
可以理解的是,由于第一推荐模型包括实体与候选搜索行为的对应关系,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系,因此,可以基于上述两种对应关系,直接生成实体与推荐信息的对应关系,以作为第三推荐模型。
例如,第三推荐模型包括的实体与推荐信息的对应关系可以为:
寻找钱婆婆的相关推荐:推荐信息A、推荐信息B、推荐信息C;
寻找千与千寻的相关推荐:推荐信息D、推荐信息E、推荐信息F;
寻找日本动漫的相关推荐:推荐信息H、推荐信息I、推荐信息J。
S102,为每组所述推荐信息生成对应的描述标签。
具体的,在获得至少两组推荐信息后,需要为其中每组推荐信息都 生成对应的描述标签。
举例说明,本发明实施例中,为每组推荐信息生成对应的描述标签的方法可以包括但不限于:
第一种:首先,根据每个目标搜索行为,确定描述关键词。然后,利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
例如,目标搜索行为包括<钱婆婆,接下来搜索千与千寻,0.59>,确定描述关键词为“千与千寻”,预设的第一标签模板可以是“关于xxx,大家都在搜”,其中xxx可以利用确定的描述关键词替换,从而可以根据描述关键词和第一标签模板生成描述标签,如“关于千与千寻,大家都在搜”。
第二种:首先,获得每组所述推荐信息对应的实体。然后,利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
例如,一组推荐信息包括“宫崎骏”、“无脸男”、“汤婆婆”,确定这一组推荐信息对应的实体为“千与千寻”,预设的第二标签模板可以是“xxx相关”,其中xxx可以利用确定的实体替换,从而可以根据每组推荐信息和第二标签模板生成描述标签,如“千与千寻相关”。
S103,在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
具体的,在获得至少两组推荐信息以及每组推荐信息对应的描述标签之后,需要在搜索结果页中对两组推荐信息和对应的描述标签进行对应显示,用以实现向用户进行搜索推荐。
举例说明,本发明实施例中,在搜索结果页中将至少两个描述标签和至少两组推荐信息进行对应显示的方法可以包括但不限于:
在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
实际应用中,可以在搜索结果页中打开的搜索结果与下一个搜索结果之间的位置,插入文本“为您推荐”,然后,将生成的描述标签显示在“为您推荐”的下方,然后在每个描述标签的右侧对应位置显示该描述标签对应的一组推荐信息。
例如,根据用户搜索的查询词和点击的某个搜索结果后,获得目标搜索行为包括A、B、C三个维度,如查询词的属性、查询词的分类、查询词的上一层需求,这些都是用户接下来希望获取的推荐信息,分别为这三个维度的目标搜索行为添加描述标签,将描述标签与对应的推荐数据分组显示给用户,从而使得用户可以快速定位感兴趣的搜索推荐,以便于用户更好地完成相关搜索任务,减少搜索成本,提升搜索效率。
实施例二
请参考图2,其为现有技术中显示的推荐信息的第一示例图,如图所示,当用户输入查询词“钱婆婆”(宫崎骏创作的日本动漫作品《千与千寻》中的一个角色)后,在与“钱婆婆”相匹配的搜索结果页中,点击了第一个搜索结果,即“钱婆婆”在百度百科的页面,在该页面用户可以了解到关于钱婆婆这个角色的基本信息。然后,由百度百科的页面返回原搜索结果页面,如图2所示,现有技术中向用户显示的推荐信息包括:钱婆婆的印章、钱婆婆和汤婆婆、千与千寻钱婆婆、钱婆婆性 格分析共四个推荐信息,推荐数量比较少,且推荐信息的类别单一。
请参考图3,其为本发明实施例所提供的显示的推荐信息的第一示例图,如图所示,为了解决现有技术中的问题,围绕“钱婆婆”这个查询词,结合用户点击了“钱婆婆”在百度百科的页面后,对用户可能的后续搜索行为进行分析,获得用户接下来三个可能感兴趣的搜索行为可以是:了解钱婆婆相关的其他信息、了解《千与千寻》动漫中的其他角色信息和了解类似的日本动漫。因此,将这三个搜索行为对应的推荐信息进行有序分组,并显示描述标签“关于钱婆婆,大家都在搜”、“关于千与千寻,大家都在搜”和“关于日本动漫,大家都在搜”,以及对应显示推荐信息,每组维度信息中可以包括4~6个推荐信息。如图3所示,在“关于钱婆婆,大家都在搜”的描述标签下推荐了“钱婆婆的印章”、“钱婆婆和汤婆婆”、“钱婆婆性格分析”;在“关于千与千寻,大家都在搜”的描述标签下推荐了“宫崎骏”、“千与千寻在线观看”、“汤婆婆”;在“关于日本动漫,大家都在搜”的描述标签下推荐了“哈尔的移动城堡”、“日本最伟大十部动画电影”、“幽灵公主”。本发明实施例所提供的技术方案,不仅为用户提供了更多可能感兴趣的推荐信息,还可以使用户能够快速定位感兴趣的搜索推荐,以便于用户更好地完成相关搜索任务,减少搜索成本,提升搜索效率。
请参考图4,其为本发明实施例所提供的显示的推荐信息的第二示例图,如图所示,若本发明实施例所提供的技术方案应用于手机时,由于手机屏幕较小,因此可以采用图4中所示出的描述标签“钱婆婆相关”、“千与千寻相关”和“日本动漫相关”。
请参考图5,其为本发明实施例所提供的显示的推荐信息的第三示 例图,如图所示,若本发明实施例所提供的技术方案应用于手机时,由于手机屏幕较小,因此可以将采用图5中的方式显示三个描述标签以及其中一个描述标签对应的一组推荐信息,可以当用户点击某一描述标签时,再显示该描述标签对应的一组推荐信息,这样,对于每个描述标签,都可以显示更多数目的推荐信息。
实施例三
请参考图6,其为现有技术中显示的推荐信息的第二示例图,如图所示,当用户输入查询词“北京住房公积金”后,在与“北京住房公积金”相匹配的搜索结果页中,点击了北京住房公积金官网的搜索结果。然后,由北京住房公积金官网的页面返回原搜索结果页面,如图6所示,向用户显示的推荐信息包括:北京住房公积金查询密码、12329用身份证查公积金共两个推荐信息,推荐数量比较少,且推荐信息的类别单一。
请参考图7,其为本发明实施例所提供的显示的推荐信息的第四示例图,如图所示,为了解决现有技术中的问题,围绕“北京住房公积金”这个查询词,结合用户点击了北京住房公积金官网的搜索结果后,对用户可能的后续搜索行为进行分析,获得用户接下来三个可能感兴趣的搜索行为可以包括:1)查询公积金额度,但在查询公积金过程中遇到问题(例如忘记密码),返回原搜索结果页继续完成公积金查询相关的下一步;2)用公积金购房,并了解公积金购房的相关问题;3)了解更多购房的相关问题。因此,将这三个搜索行为对应的推荐信息进行有序分组,并显示描述标签“关于该网站登陆问题,大家都在搜”、“关于公积金购房问题,大家都在搜”和“关于购房问题,大家都在搜”,以及对应显示推荐信息,每组维度信息中可以包括2~4个推荐信息。如图7所示, 在“关于该网站登陆问题,大家都在搜”的描述标签下推荐了“北京住房公积金查询密码”、“北京住房公积金重置密码”;在“关于公积金购房问题,大家都在搜”的描述标签下推荐了“北京公积金提取”、“北京公积金购房流程”、“公积金最高可贷多少”;在“关于购房问题,大家都在搜”的描述标签下推荐了“组合贷款流程”、“便宜的二手房”、“靠谱的购房中介”和“总价200万内的房子”。本发明实施例所提供的技术方案,不仅为用户提供了更多可能感兴趣的推荐信息,还可以使用户能够快速定位感兴趣的搜索推荐,以便于用户更好地完成相关搜索任务,减少搜索成本,提升搜索效率。
本发明实施例进一步给出实现上述方法实施例中各步骤及方法的装置实施例。
请参考图8,其为本发明实施例所提供的搜索推荐装置的功能方块图。如图所示,该装置包括:
信息获取模块81,用于若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;
标签获取模块82,用于为每组所述推荐信息生成对应的描述标签;
推荐显示模块83,用于在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
在一个具体的实现过程中,所述信息获取模块81,具体用于:
根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为;
根据每个所述目标搜索行为,获得对应的一组推荐信息。
在一个具体的实现过程中,所述信息获取模块81用于根据所述搜索 结果所匹配的实体,获得用户的至少两个目标搜索行为时,具体用于:
将与所述搜索结果相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出;
其中,所述第一推荐模型包括实体与候选搜索行为的对应关系。
在一个具体的实现过程中,所述信息获取模块81用于根据每个所述目标搜索行为,获得对应的一组推荐信息时,具体用于:
将每个所述目标搜索行为输入第二推荐模型,以使得所述第二推荐模型根据输入的目标搜索行为,获得并输出对应的一组推荐信息;
其中,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系。
在一个具体的实现过程中,所述标签获取模块82,具体用于:
根据每个所述目标搜索行为,确定描述关键词;
利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
在一个具体的实现过程中,所述标签获取模块82,具体用于:
获得每组所述推荐信息对应的实体;
利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
在一个具体的实现过程中,所述推荐显示模块83,具体用于:
在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
由于本实施例中的各单元能够执行图1所示的方法,本实施例未详 细描述的部分,可参考对图1的相关说明。
本发明实施例的技术方案具有以下有益效果:
本发明实施例中,若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;从而,为每组所述推荐信息生成对应的描述标签;进而,在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
根据本发明实施例提供的技术方案,可以在用户点击某搜索结果后,在搜索结果页面上向用户提供多组推荐信息且推荐信息分组显示,并为每组推荐信息添加描述标签,以便于用户根据描述标签和推荐信息的分组,能够及时获得所需要的推荐信息,与现有技术中用户人工筛选推荐信息和提供单一维度的推荐信息相比,本发明实施例所提供的技术方案,提高了搜索结果页中推荐信息的多样性以及获取信息的效率。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统,装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
在本发明所提供的几个实施例中,应该理解到,所揭露的系统,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述单元的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式,例如,多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略,或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本发明各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元,可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中,包括若干指令用以使得一台计算机装置(可以是个人计算机,服务器,或者网络装置等)或处理器(Processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本发明保护的范围之内。

Claims (16)

  1. 一种搜索推荐方法,其特征在于,所述方法包括:
    若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;
    为每组所述推荐信息生成对应的描述标签;
    在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
  2. 根据权利要求1所述的方法,其特征在于,所述获得至少两组推荐信息,包括:
    根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为;
    根据每个所述目标搜索行为,获得对应的一组推荐信息。
  3. 根据权利要求2所述的方法,其特征在于,所述根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为,包括:
    将与所述搜索结果相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出;
    其中,所述第一推荐模型包括实体与候选搜索行为的对应关系。
  4. 根据权利要求2所述的方法,其特征在于,根据每个所述目标搜索行为,获得对应的一组推荐信息,包括:
    将每个所述目标搜索行为输入第二推荐模型,以使得所述第二推荐模型根据输入的目标搜索行为,获得并输出对应的一组推荐信息;
    其中,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系。
  5. 根据权利要求2所述的方法,其特征在于,为每组所述推荐信息生成对应的描述标签,包括:
    根据每个所述目标搜索行为,确定描述关键词;
    利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
  6. 根据权利要求1或4所述的方法,其特征在于,为每组所述推荐信息生成对应的描述标签,包括:
    获得每组所述推荐信息对应的实体;
    利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
  7. 根据权利要求1所述的方法,其特征在于,在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示,包括:
    在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
  8. 一种搜索推荐装置,其特征在于,所述装置包括:
    信息获取模块,用于若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;
    标签获取模块,用于为每组所述推荐信息生成对应的描述标签;
    推荐显示模块,用于在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
  9. 根据权利要求8所述的装置,其特征在于,所述信息获取模块, 具体用于:
    根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为;
    根据每个所述目标搜索行为,获得对应的一组推荐信息。
  10. 根据权利要求9所述的装置,其特征在于,所述信息获取模块用于根据所述搜索结果所匹配的实体,获得用户的至少两个目标搜索行为时,具体用于:
    将与所述搜索结果相匹配的实体输入第一推荐模型,以使得所述第一推荐模型根据所述实体获得与所述实体相匹配的至少两个候选搜索行为,以作为所述至少两个目标搜索行为并输出;
    其中,所述第一推荐模型包括实体与候选搜索行为的对应关系。
  11. 根据权利要求9所述的装置,其特征在于,所述信息获取模块用于根据每个所述目标搜索行为,获得对应的一组推荐信息时,具体用于:
    将每个所述目标搜索行为输入第二推荐模型,以使得所述第二推荐模型根据输入的目标搜索行为,获得并输出对应的一组推荐信息;
    其中,所述第二推荐模型包括候选搜索行为与推荐信息的对应关系。
  12. 根据权利要求9所述的装置,其特征在于,所述标签获取模块,具体用于:
    根据每个所述目标搜索行为,确定描述关键词;
    利用所述描述关键词和预设的第一标签模板,为每组所述推荐信息生成对应的描述标签。
  13. 根据权利要求8或11所述的装置,其特征在于,所述标签获 取模块,具体用于:
    获得每组所述推荐信息对应的实体;
    利用所述实体和预设的第二标签模板,为每组所述推荐信息生成对应的描述标签。
  14. 根据权利要求8所述的装置,其特征在于,所述推荐显示模块,具体用于:
    在所述搜索结果页中打开的所述搜索结果与下一个搜索结果之间,显示至少两个所述描述标签,并在每个所述描述标签的指定区域内显示对应的一组推荐信息。
  15. 一种设备,包括
    一个或者多个处理器;
    存储器;
    一个或者多个程序,所述一个或者多个程序存储在所述存储器中,当被所述一个或者多个处理器执行时:
    若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返回所述搜索结果页,获得至少两组推荐信息;
    为每组所述推荐信息生成对应的描述标签;
    在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
  16. 一种计算机存储介质,所述计算机存储介质被编码有计算机程序,所述程序在被一个或多个计算机执行时,使得所述一个或多个计算机执行如下操作:
    若在搜索结果页中打开一个搜索结果,并由所述搜索结果的页面返 回所述搜索结果页,获得至少两组推荐信息;
    为每组所述推荐信息生成对应的描述标签;
    在所述搜索结果页中将至少两个所述描述标签和至少两组所述推荐信息进行对应显示。
PCT/CN2016/089483 2015-11-25 2016-07-08 一种搜索推荐方法、装置、设备及计算机存储介质 WO2017088496A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510828621.3 2015-11-25
CN201510828621.3A CN105512180B (zh) 2015-11-25 2015-11-25 一种搜索推荐方法及装置

Publications (1)

Publication Number Publication Date
WO2017088496A1 true WO2017088496A1 (zh) 2017-06-01

Family

ID=55720162

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/089483 WO2017088496A1 (zh) 2015-11-25 2016-07-08 一种搜索推荐方法、装置、设备及计算机存储介质

Country Status (2)

Country Link
CN (1) CN105512180B (zh)
WO (1) WO2017088496A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111078989A (zh) * 2018-10-18 2020-04-28 阿里巴巴集团控股有限公司 一种应用程序的推荐方法、装置、及电子设备

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105512180B (zh) * 2015-11-25 2019-07-23 百度在线网络技术(北京)有限公司 一种搜索推荐方法及装置
CN106250421A (zh) * 2016-07-25 2016-12-21 深圳市金立通信设备有限公司 一种拍摄处理的方法及终端
CN106294730A (zh) * 2016-08-09 2017-01-04 百度在线网络技术(北京)有限公司 资讯信息的推荐方法及装置
CN106557547A (zh) * 2016-10-24 2017-04-05 百度在线网络技术(北京)有限公司 基于人工智能的提供资讯内容的方法及装置
CN108205572A (zh) * 2016-12-20 2018-06-26 百度在线网络技术(北京)有限公司 一种搜索方法、装置及设备
CN107247609B (zh) * 2017-06-30 2020-07-14 江西博瑞彤芸科技有限公司 字库维护管理方法
CN112860155B (zh) * 2020-07-10 2022-06-03 浙江口碑网络技术有限公司 一种搜索引导的处理方法及装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110202520A1 (en) * 2008-11-26 2011-08-18 Yahoo! Inc. Navigation assistance for search engines
CN102693271A (zh) * 2012-03-06 2012-09-26 奇智软件(北京)有限公司 一种网络信息推荐方法及系统
US8719244B1 (en) * 2005-03-23 2014-05-06 Google Inc. Methods and systems for retrieval of information items and associated sentence fragments
CN103870461A (zh) * 2012-12-10 2014-06-18 腾讯科技(深圳)有限公司 主题推荐方法、装置和服务器
CN103914553A (zh) * 2014-04-14 2014-07-09 百度在线网络技术(北京)有限公司 搜索方法和搜索引擎
CN105512180A (zh) * 2015-11-25 2016-04-20 百度在线网络技术(北京)有限公司 一种搜索推荐方法及装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102567408B (zh) * 2010-12-31 2014-06-04 阿里巴巴集团控股有限公司 推荐搜索关键词的方法和装置
US20140358916A1 (en) * 2013-05-29 2014-12-04 Microsoft Corporation Personalized prioritization of integrated search results
CN104239495B (zh) * 2014-09-09 2018-06-05 百度在线网络技术(北京)有限公司 搜索方法和搜索装置
CN104281699B (zh) * 2014-10-15 2017-11-17 百度在线网络技术(北京)有限公司 搜索推荐方法及装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8719244B1 (en) * 2005-03-23 2014-05-06 Google Inc. Methods and systems for retrieval of information items and associated sentence fragments
US20110202520A1 (en) * 2008-11-26 2011-08-18 Yahoo! Inc. Navigation assistance for search engines
CN102693271A (zh) * 2012-03-06 2012-09-26 奇智软件(北京)有限公司 一种网络信息推荐方法及系统
CN103870461A (zh) * 2012-12-10 2014-06-18 腾讯科技(深圳)有限公司 主题推荐方法、装置和服务器
CN103914553A (zh) * 2014-04-14 2014-07-09 百度在线网络技术(北京)有限公司 搜索方法和搜索引擎
CN105512180A (zh) * 2015-11-25 2016-04-20 百度在线网络技术(北京)有限公司 一种搜索推荐方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111078989A (zh) * 2018-10-18 2020-04-28 阿里巴巴集团控股有限公司 一种应用程序的推荐方法、装置、及电子设备
CN111078989B (zh) * 2018-10-18 2024-03-22 阿里巴巴集团控股有限公司 一种应用程序的推荐方法、装置、及电子设备

Also Published As

Publication number Publication date
CN105512180B (zh) 2019-07-23
CN105512180A (zh) 2016-04-20

Similar Documents

Publication Publication Date Title
WO2017088496A1 (zh) 一种搜索推荐方法、装置、设备及计算机存储介质
KR101511050B1 (ko) 상품 정보를 제공하고 표시하는 방법, 장치, 시스템 및 컴퓨터 프로그램
JP6022056B2 (ja) 検索結果の生成
KR102035435B1 (ko) 검색을 시드하는 시스템 및 방법
TW201812689A (zh) 惡意地址/惡意訂單的識別系統、方法及裝置
US9659067B2 (en) Providing a search results document that includes a user interface for performing an action in connection with a web page identified in the search results document
CN107480158A (zh) 基于相似性得分评估内容项目与图像的匹配的方法和系统
CN107784092A (zh) 一种推荐热词的方法、服务器及计算机可读介质
JP6646931B2 (ja) 推薦情報を提供するための方法および装置
CN107222526B (zh) 推送推广信息的方法、装置、设备和计算机存储介质
US8655737B1 (en) Brand name synonymy
WO2015179756A1 (en) Systems and methods for customizing search results and recommendations
WO2020114145A1 (zh) 内容推送方法及存储介质、计算机设备
WO2015135110A1 (en) Systems and methods for keyword suggestion
US11860955B2 (en) Method and system for providing alternative result for an online search previously with no result
TW201401088A (zh) 搜索方法和裝置
CN110674620A (zh) 目标文案生成方法、装置、介质及电子设备
CN110766486A (zh) 确定物品类目的方法和装置
US10255246B1 (en) Systems and methods for providing a searchable concept network
CN110827101A (zh) 一种店铺推荐的方法和装置
CN111144122A (zh) 评价处理方法、装置和计算机系统及介质
CN114862480A (zh) 广告投放定向方法及其装置、设备、介质、产品
TWI820489B (zh) 關鍵字管理系統
CN113378065B (zh) 一种基于滑动谱分解确定内容多样性的方法和选定内容的方法
CN113297471A (zh) 生成数据对象标签、数据对象搜索方法、装置及电子设备

Legal Events

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

Ref document number: 16867723

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16867723

Country of ref document: EP

Kind code of ref document: A1