WO2023231288A1 - 搜索结果的排序方法、装置、电子设备和存储介质 - Google Patents

搜索结果的排序方法、装置、电子设备和存储介质 Download PDF

Info

Publication number
WO2023231288A1
WO2023231288A1 PCT/CN2022/128370 CN2022128370W WO2023231288A1 WO 2023231288 A1 WO2023231288 A1 WO 2023231288A1 CN 2022128370 W CN2022128370 W CN 2022128370W WO 2023231288 A1 WO2023231288 A1 WO 2023231288A1
Authority
WO
WIPO (PCT)
Prior art keywords
search
search result
search results
score
historical
Prior art date
Application number
PCT/CN2022/128370
Other languages
English (en)
French (fr)
Inventor
刘伟
黄鑫安
覃黎飞
陈海歌
杨昱
李婷
卜建国
林赛群
Original Assignee
北京百度网讯科技有限公司
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 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Priority to EP22931264.0A priority Critical patent/EP4307136A1/en
Publication of WO2023231288A1 publication Critical patent/WO2023231288A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/953Querying, e.g. by the use of web search engines
    • G06F16/9538Presentation of query results
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/245Query processing
    • G06F16/2457Query processing with adaptation to user needs
    • G06F16/24578Query processing with adaptation to user needs using ranking
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21Design, administration or maintenance of databases
    • G06F16/215Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present disclosure relates to the field of computer technology, specifically to the field of artificial intelligence technology such as big data, and in particular to a sorting method, device, electronic device and storage medium for search results.
  • the present disclosure aims to solve one of the technical problems in the related art, at least to a certain extent.
  • a method for sorting search results including:
  • the at least one first search result is push information
  • the initial ranking is adjusted in descending order of the first scores to obtain an updated ranking of the plurality of first search results.
  • a sorting device for search results including:
  • a first acquisition module configured to acquire a plurality of first search results corresponding to the search request and an initial ranking of the plurality of first search results
  • a first determination module configured to determine the type and style of the at least one first search result when the at least one first search result is push information
  • a second determination module configured to determine the first score corresponding to each first search result based on the type and style of each first search result in the at least one first search result based on a preset mapping relationship
  • the second acquisition module is configured to adjust the initial ranking in descending order of the first scores to obtain an updated ranking of the plurality of first search results.
  • the third embodiment of the present disclosure provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the present disclosure is implemented.
  • the fourth embodiment of the present disclosure provides a non-transitory computer-readable storage medium that stores a computer program.
  • the computer program is executed by a processor, the method proposed in the first embodiment of the present disclosure is implemented.
  • the fifth embodiment of the present disclosure provides a computer program product.
  • instructions in the computer program product are executed by a processor, the method proposed in the first embodiment of the present disclosure is executed.
  • multiple first search results corresponding to the search request and the initial ordering of the multiple first search results are first obtained, and then when at least one first search result is push information, it is determined that the at least one first search result is push information.
  • the type and style of a first search result and then based on the preset mapping relationship, according to the type and style of each first search result in the at least one first search result, determine the corresponding number of each first search result. first score, and then adjust the initial ranking in descending order of the first score to obtain an updated ranking of the plurality of first search results.
  • the first search results that are push information can be scored again, and then the initial ranking of each first search result can be updated. and adjustments, so that the final sorting can not only meet the user's search needs, but also tap the information value of the search results, realize information push, and improve the value brought by the sorting results.
  • Figure 1 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure
  • Figure 2 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure
  • Figure 3 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure
  • Figure 4 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure
  • Figure 5 is a schematic structural diagram of a search result sorting device provided by some embodiments of the present disclosure.
  • FIG. 6 is a block diagram of an electronic device used to implement a method for sorting search results according to an embodiment of the present disclosure.
  • Artificial intelligence is the study of using computers to simulate certain human thinking processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.). It has both hardware-level technology and software-level technology. Artificial intelligence hardware technology generally includes sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing and other technologies; artificial intelligence software technology mainly includes computer vision technology, speech recognition technology, natural language processing technology and machine learning, deep Learning, big data processing technology, knowledge graph technology and other major directions.
  • Big data processing refers to the process of analyzing and processing huge amounts of data using artificial intelligence. Big data can be summarized as five Vs: Volume, Velocity, and Variety. ), value (Value), authenticity (Veracity).
  • the present disclosure provides a method for sorting search results.
  • the method can be executed by a sorting device for search results provided by the present disclosure, or can be executed by an electronic device provided by the present disclosure.
  • the electronic device can include but is not limited to a desktop computer, Tablets, cloud devices, mobile devices, personal digital assistants, and other hardware devices with various operating systems, touch screens and/or displays, or, alternatively, servers.
  • the search system provided by the present disclosure is used below to perform a sorting method of search results provided by the present disclosure, which is not intended to limit the present disclosure.
  • Figure 1 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure.
  • the sorting method of the search results can include the following steps:
  • Step 101 Obtain multiple first search results corresponding to the search request and an initial ranking of the multiple first search results.
  • the search request may be generated by the search system based on any search text input by the user, or may be generated based on any search item clicked by the user, which is not limited here.
  • the search system can generate a search request corresponding to "Apple” and display corresponding search results to the user based on the search request.
  • the first search result may be a search result obtained by a search system (search engine) based on a user's search request.
  • search engine search engine
  • the first search result may be a search engine results page (SERP), which refers to the result page that the search system responds to the current search request.
  • SERP search engine results page
  • the search system can first perform data cleaning on each first search result to filter out abnormal data, and then based on the relationship between each first search result and the search request.
  • the first search results are ranked based on the degree of correlation between them, and an initial ranking is generated.
  • the correlation between each first search result and the search request can be calculated based on a machine learning algorithm, such as the TF/IDF algorithm (Term Frequency&Inverse Document Frequency), which is not limited here.
  • TF/IDF algorithm Term Frequency&Inverse Document Frequency
  • Y-Z-X can be determined as The initial sorting is not limited here.
  • the search system may also determine multiple first search results based on the relevance of each first search result to the search request and the click volume of each first search result within a preset period. Initial ranking of first search results.
  • the preset period can be a period of a specified period, for example, it can be a historical month, or a historical week, which is not limited here. For example, if the current time is April 1, the time between March 1 and April 1 can be used as the default period.
  • the preset period in the present disclosure can be set to a recent historical time period, such as a time within one week from the current time.
  • the first search results corresponding to the current search request Q are X, Y, and Z respectively, and the correlations between X, Y, and Z and the search request Q are 0.8 and 0.6 respectively.
  • the example is only a schematic illustration of the present disclosure and does not constitute a limitation on the present disclosure.
  • the initial ranking of the plurality of first search results can be made to better match the user's search intention, and the user can more conveniently obtain content with popular value.
  • Step 102 When at least one first search result is push information, determine the type and style of at least one first search result.
  • the push information can be information with commercial value, such as information with high search frequency, information with high revenue conversion rate, etc., which are not limited here.
  • the type can be a comprehensive page, a list page, a content page, etc., and is not limited here.
  • the style can be pictures, videos, cards, etc., which are not limited here.
  • each first search result can be input into a neural network model generated by pre-training to label the value corresponding to each first search result, for example, it can be "high value”, “medium value” ",”low value”. Then, the first search result marked as "high value” is determined as push information.
  • At least one first search result corresponding to the push information can be sorted. It should be noted that when ranking at least one first search result, the first search result can be measured based on multiple dimensions, such as quality, timeliness, and relevance, which are not limited here. In the present disclosure, the push information can be sorted according to the page characteristics of the at least one first search result.
  • page characteristics may include type, style, etc.
  • the present disclosure can judge the type and style of the first search result corresponding to the current push information, and then proceed based on this. Evaluate.
  • Step 103 Based on a preset mapping relationship, determine the first score corresponding to each first search result according to the type and style of each first search result in at least one first search result.
  • the first score may be a score for each type and style of page. Specifically, the first scores of pages of the same type and style are the same, and the first scores of pages of different types and styles may be the same or different.
  • the present disclosure can pre-set the score corresponding to each type and style based on the number of clicks or revenue conversion rate of users on pages of various types and styles in historical data, and use the score to represent the user's liking for the page content and the High or low value.
  • a mapping relationship table may be set in advance, and the first score corresponding to the type and style of the page content may be preset in the mapping relationship table. Furthermore, after determining each first search result corresponding to the current push information, the search system can determine the first score corresponding to each first search result according to the current type and style of each first search result.
  • Step 104 Adjust the initial ranking in descending order of the first scores to obtain an updated ranking of multiple first search results.
  • the first score of the at least one first search result can be determined, and the at least one first search result can be ranked according to the order of the first score from large to small. Sort and further adjust the initial sorting.
  • the initial sorting is r1, r2, r3, r4, r5, r6, r7.
  • r1, r3, and r6 are push information
  • the first scores corresponding to r1, r3, and r6 are 2, 5, and 4 respectively.
  • r1, r3, and r6 can be reordered into r3, r6, and r1, and the initial The sorting is updated, and the updated sorting is obtained as r3, r2, r6, r4, r5, r1, r7, which is not limited here.
  • each first search result corresponding to the push information can also be placed at the top, that is, the updated sorting can be obtained as r3, r6, r1, r2, r4, r5, r7, which is not limited here.
  • multiple first search results corresponding to the search request and the initial ordering of the multiple first search results are first obtained, and then when at least one first search result is push information, it is determined that the at least one first search result is push information.
  • the type and style of a first search result and then based on the preset mapping relationship, according to the type and style of each first search result in the at least one first search result, determine the corresponding number of each first search result. first score, and then adjust the initial ranking in descending order of the first score to obtain an updated ranking of the plurality of first search results.
  • the first search results that are push information can be scored again, and then the initial ranking of each first search result can be updated. and adjustments, so that the final sorting can not only meet the user's search needs, but also tap the information value of the search results, realize information push, and improve the value brought by the sorting results.
  • Figure 2 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure.
  • the sorting method of the search results may include the following steps:
  • Step 201 Obtain multiple first search results corresponding to the search request.
  • step 201 may refer to the above embodiment, and will not be described again here.
  • Step 202 Determine at least one vertical category tag currently mapped by the search request.
  • vertical category labels can be vertical domain classification labels. It should be noted that a search request may be mapped to one or more different vertical tags, and vertical tags can be used to characterize the field or industry to which the current search content belongs.
  • the vertical tags mapped by the search text can be ⁇ poetry ⁇ , ⁇ road name ⁇ , ⁇ flower ⁇ , ⁇ award ⁇ , ⁇ movie ⁇ , Wait, there is no limit here.
  • the vertical tags mapped by the search text can be ⁇ cooking school ⁇ , ⁇ English training school ⁇ , ⁇ direction ⁇ , etc., which are not limited here.
  • Step 203 Calculate the second score of each vertical category label in at least one vertical category label based on the preset value index.
  • CTR click-through rate
  • CVR revenue conversion rate
  • search frequency average click price
  • CPC average click price
  • CPA average conversion price
  • CPM price per thousand impressions
  • the second score may be a score for the vertical category label, used to represent the value of the vertical category label. That is to say, the higher the second score, the greater the value of the corresponding vertical category label.
  • the search system can obtain various types of data corresponding to the value indicators based on the preset value indicators, and then perform data cleaning to filter out abnormal data, such as low-quality, cheating, empty, expired and other data.
  • multiple pages corresponding to each vertical tag can be selected, and based on the preset value indicators, the scores corresponding to the value indicators of each page are obtained, and then based on the scores of each page on each value indicator, and The weight corresponding to each value indicator is calculated, and the score corresponding to each vertical category label is calculated.
  • value_score sigmoid( ⁇ w*value_info)
  • value_score is the second score corresponding to the vertical label
  • value_info is the score corresponding to each value indicator
  • w is the weight corresponding to each value indicator
  • sigmoid is the activation function
  • Step 204 Determine the correlation between each first search result and each vertical category tag.
  • the correlation between each first search result and each vertical category tag can be calculated based on a machine learning algorithm, such as the TF/IDF algorithm (Term Frequency&Inverse Document Frequency), which is not limited here.
  • TF/IDF algorithm Term Frequency&Inverse Document Frequency
  • Step 205 Determine the third score of each first search result based on the correlation between each first search result and each vertical category tag and the second score of the vertical category tag.
  • first search results namely A, B, C, D, and E
  • vertical category labels namely x, y, and z
  • the second scores of the labels are 20, 15, and 10 respectively.
  • the second score and relevance of each first search result on each vertical category label can be multiplied, and finally the multiplied results are added to calculate the third score of each first search result.
  • Step 206 Determine the initial ranking of the plurality of first search results based on the third scores respectively corresponding to the plurality of first search results.
  • A, B, C, D, and E can be ranked For A, E, C, B, D.
  • Step 207 Determine the historical search frequency and historical revenue conversion rate corresponding to each first search result.
  • the revenue conversion rate can be the conversion rate of users registering, activating or becoming paying users by clicking on any search results within a specified historical period.
  • the first threshold can be the threshold of historical search frequency. If the historical search frequency is greater than the first threshold, it means that the first search result is relatively popular and the number of searches by the user is very large.
  • the size of the first threshold can be set based on actual experience. .
  • the second threshold can be the threshold of the historical income conversion rate. If the historical income conversion rate is greater than the second threshold, it means that the conversion rate of the first search result is relatively high and the income it brings is relatively large.
  • the size of the second threshold can be Set based on actual experience.
  • any historical search result has a high historical search frequency or a high revenue conversion rate, it means that the historical search result is more popular with users and is content that users tend to choose.
  • push information can be information that has certain commercial value and that users are more likely to choose. By pushing push information to users, users can be exposed to content with more commercial value.
  • the first search results in each first search result whose historical search frequency is greater than the first threshold can be used as push information, or the first search results in each first search result whose historical revenue conversion rate is greater than the second threshold can also be used as push information.
  • the first search results may be used as push information, or the first search results whose historical revenue conversion rate is greater than the second threshold and whose historical search frequency is greater than the first threshold may be used as push information, which is not limited here.
  • Step 209 When at least one first search result is push information, determine the type and style of at least one first search result.
  • Step 210 Based on the preset mapping relationship, determine the first score corresponding to each first search result according to the type and style of each first search result in at least one first search result.
  • Step 211 Adjust the initial ranking in descending order of the first scores to obtain an updated ranking of multiple first search results.
  • steps 209, 210, and 211 can refer to any of the above embodiments, and will not be described again here.
  • multiple first search results corresponding to the search request are first obtained, and then at least one vertical category tag currently mapped by the search request is determined, and then based on the preset value indicator, each of the at least one vertical category tag is calculated.
  • the second score of the vertical category label and then determine the correlation between each first search result and each vertical category label, and then based on the correlation between each first search result and each vertical category label, and the third score of the vertical category label Second score, determine the third score of each first search result, and then determine the initial ranking of the multiple first search results based on the third scores corresponding to the multiple first search results, and then determine the initial ranking of each first search result.
  • the historical search frequency and historical revenue conversion rate corresponding to the result, and then the first search result with the historical search frequency greater than the first threshold and/or the historical revenue conversion rate greater than the second threshold is determined as push information, and then in at least one first search
  • the result is push information
  • determine the type and style of at least one first search result and then determine each first search result based on the preset mapping relationship based on the type and style of each first search result in the at least one first search result.
  • the first score corresponding to the first search result is finally adjusted in order of the first score from large to small to obtain an updated sorting of multiple first search results.
  • the initial ranking is determined based on the correlation between each first search result and the vertical category tag and the second score of the vertical category tag, the initial ranking can take into account the correlation between the first search result and the current search request.
  • the value of each vertical tag mapped by the search request is also taken into consideration, so that the current initial ranking can meet the user's search needs and have higher commercial value.
  • Figure 3 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure.
  • the sorting method of the search results may include the following steps:
  • Step 302 When at least one first search result is push information, determine the type and style of the at least one first search result.
  • the specified historical period can be a pre-selected time period, for example, it can be a time period within one week from the current date, which is not limited here. It should be noted that the end point of the specified historical period can be the current date, that is to say, the specified historical period is a time period connected to the current date, so that each historical search result of the current historical period can represent the current user's choice tendency. and personal intention, that is, taking into account the impact of timeliness on ranking.
  • the fourth score may be a score for historical search results, used to represent the value of historical search results. In other words, the higher the fourth score, the greater the value of the corresponding historical search results.
  • value_score sigmoid( ⁇ w*value_info)
  • value_score is the fourth score corresponding to the historical search results
  • value_info is the score corresponding to each value indicator
  • w is the weight corresponding to each value indicator
  • sigmoid is the activation function
  • each historical search result has a corresponding type and style.
  • the type can be a content page, a list page, a comprehensive page, etc., and is not limited here.
  • the style can be pictures, videos, cards, etc., which are not limited here.
  • each historical search result After obtaining each historical search result, each historical search result can be classified according to type and style, so that the mean value of the fourth score corresponding to each category, that is, each historical search result of the same type and style can be determined.
  • the fourth score average of each historical search result with the same type and style can be determined as the first score corresponding to that type and style. That is to say, a relationship between the type and style and the first score can be established. mapping relationship.
  • Step 307 Adjust the initial ranking in descending order of the first scores to obtain an updated ranking of multiple first search results.
  • each first search result is Q1, Q2, Q3, Q4, Q5, Q6, Q7
  • the corresponding arrangement positions of Q1, Q2, Q3, Q4, Q5, Q6, Q7 can be determined, for example, Q1 , Q2, Q3, and Q4 are arranged in the first row of the first page, the second row of the first page, the third row of the first page, and the fourth row of the first page.
  • Q5, Q6, and Q7 are arranged respectively in The first line of the second page, the second line of the second page, and the third line of the second page are not limited here.
  • Step 309 Insert the second search result into the designated display position in the arrangement position, where the second search result is a predetermined search result with designated page characteristics or designated page content.
  • the designated display position may be a pre-specified arrangement position, such as the first position, the second position, or the third position, which is not limited here. There can be one or more designated display locations.
  • the second search result may be inserted into the specified display position in the arrangement position.
  • the search system can insert the second search result in front of the first search result that was originally the first position, and update the current ranking.
  • the multiple second search results can be inserted into each designated display position in the arrangement position in a preset order.
  • the three second search results can be arranged in the first position, the second position, and the third position in a preset order, and the current sorting can be performed. renew.
  • multiple first search results corresponding to the search request and the initial ordering of the multiple first search results are first obtained, and then when at least one first search result is push information, it is determined that the at least one first search result is push information.
  • the type and style of a first search result then obtain each historical search result in a specified historical period, and finally determine the fourth score corresponding to each historical search result based on the preset value indicator, and then determine the fourth score corresponding to the historical search result based on the same type and style.
  • the fourth score average of each of the historical search results determine the mapping relationship between each type and style and the first score, and then based on the preset mapping relationship, according to the first search result in the at least one first search result Type and style, determine the first score corresponding to each first search result, and then adjust the initial sorting in order of the first score from large to small to obtain the updated sorting of multiple first search results, and then determine The current arrangement position of each first search result, and finally inserting the second search result into the designated display position in the arrangement position. Therefore, after sorting the first search results, the second search results in other dimensions can be combined with the first search results, thereby increasing the value of the sorted results and better meeting the user's search needs.
  • Figure 4 is a schematic flowchart of a method for sorting search results provided by some embodiments of the present disclosure.
  • the sorting method of the search results may include the following steps:
  • Step 401 Obtain multiple first search results corresponding to the search request and attribute information of the device that initiated the search request.
  • the search request may be generated by the search system based on any search text input by the user, or may be generated based on any search item clicked by the user, which is not limited here.
  • the search system can generate a search request corresponding to "Apple” and display corresponding search results to the user based on the search request.
  • the first search result may be a search result obtained by a search system (search engine) based on a user's search request.
  • search engine search engine
  • the first search result may be a search engine results page (SERP), which refers to the result page that the search system responds to the current search request.
  • SERP search engine results page
  • the device attribute information that initiates the search request may be device-related information or device user-related information, such as the type of device, the type of network used by the device, and information about the user to which the device belongs, which is not limited here.
  • the device attribute information includes at least one of the following: a type of device, a network type used by the device, and attribute information of a user to which the device belongs.
  • the type of device can be a computer, a mobile phone, a tablet, etc., and is not limited here.
  • the network type used by the device can be a local area network, a metropolitan area network, a wide area network, and the Internet, which are not limited here.
  • the user's attribute information can be the user's gender, age, occupation, device usage time, page operation record information, etc., which are not limited here.
  • Step 402 Determine the reference ordering associated with the device attribute information according to a preset mapping relationship.
  • the device that initiates the search request will affect the user's habit of browsing the first search results to a certain extent.
  • computer devices are more suitable for playing video-type pages than mobile devices
  • tablets are more suitable for playing video-type pages than mobile devices.
  • users of devices of different ages usually choose different page types.
  • the types of pages that girls like to browse are also different from the types of pages that boys like to browse.
  • its corresponding reference ordering can be U, V, and Z.
  • device attribute information B its corresponding reference ordering can be U, Z, V.
  • device attribute information C its corresponding reference ordering can be The sorting can be Z, V, or U, and is not limited here.
  • U, V, and Z can be different page features, such as a comprehensive long video page, a card list page, and a short video content page, which are not limited here.
  • the search system can sort each first search result according to the page characteristics of each first search result and the reference ranking corresponding to the current page characteristics, thereby determining the initial ranking of multiple first search results. .
  • Step 406 Adjust the initial ranking in descending order of the first scores to obtain an updated ranking of the plurality of first search results.
  • steps 404, 405, and 406 may refer to the above embodiment, and will not be described again here.
  • the first obtaining module 510 is used to obtain a plurality of first search results corresponding to the search request and an initial ranking of the plurality of first search results.
  • the first acquisition module 510 is specifically used to:
  • the first acquisition module 510 is also used to:
  • the first search results whose historical search frequency is greater than the first threshold and/or whose historical revenue conversion rate is greater than the second threshold are determined as push information.
  • the second determination module 530 is also used to:
  • An inserting unit is configured to insert a second search result into a designated display position in the arrangement position, wherein the second search result is a predetermined search result with designated page characteristics or designated page content.
  • multiple second search results are inserted into each of the designated display positions in the arrangement positions in a preset order.
  • multiple first search results corresponding to the search request and the initial ordering of the multiple first search results are first obtained, and then when at least one first search result is push information, it is determined that the at least one first search result is push information.
  • the type and style of a first search result and then based on the preset mapping relationship, according to the type and style of each first search result in the at least one first search result, determine the corresponding number of each first search result. first score, and then adjust the initial ranking in descending order of the first score to obtain an updated ranking of the plurality of first search results.
  • FIG. 6 shows a schematic block diagram of an example electronic device 500 that may be used to implement embodiments of the present disclosure.
  • Electronic devices are intended to refer to various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.
  • Electronic devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices.
  • the components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit implementations of the disclosure described and/or claimed herein.
  • the device 500 includes a computing unit 501 that can execute according to a computer program stored in a read-only memory (ROM) 502 or loaded from a storage unit 508 into a random access memory (RAM) 503 Various appropriate actions and treatments. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored.
  • Computing unit 501, ROM 502 and RAM 503 are connected to each other via bus 504.
  • An input/output (I/O) interface 505 is also connected to bus 504.
  • I/O interface 505 Multiple components in the device 500 are connected to the I/O interface 505, including: input unit 506, such as a keyboard, mouse, etc.; output unit 507, such as various types of displays, speakers, etc.; storage unit 508, such as a magnetic disk, optical disk, etc. ; and communication unit 509, such as a network card, modem, wireless communication transceiver, etc.
  • the communication unit 509 allows the device 500 to exchange information/data with other devices through computer networks such as the Internet and/or various telecommunications networks.
  • Computing unit 501 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc.
  • the computing unit 501 performs various methods and processes described above, such as the training method of the graph processing network model.
  • the training method of the graph processing network model may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as the storage unit 508.
  • part or all of the computer program may be loaded and/or installed onto device 500 via ROM 502 and/or communication unit 509.
  • the computer program When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the training method of the graph processing network model described above may be performed.
  • the computing unit 501 may be configured to perform the training method of the graph processing network model in any other suitable manner (eg, by means of firmware).
  • Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or a combination thereof.
  • FPGAs field programmable gate arrays
  • ASICs application specific integrated circuits
  • ASSPs application specific standard products
  • SOC system
  • CPLD load programmable logic device
  • computer hardware firmware, software, and/or a combination thereof.
  • These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor
  • the processor which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • An output device may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • An output device may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
  • Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions specified in the flowcharts and/or block diagrams/ The operation is implemented.
  • the program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
  • a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.
  • the machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing.
  • machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, 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 above.
  • RAM random access memory
  • ROM read only memory
  • EPROM or flash memory erasable programmable read only memory
  • CD-ROM portable compact disk read-only memory
  • magnetic storage device or any suitable combination of the above.
  • the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer.
  • a display device eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor
  • a keyboard and pointing device eg, a mouse or a trackball
  • Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.
  • the systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., A user's computer having a graphical user interface or web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system.
  • the components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), the Internet, and blockchain networks.
  • Computer systems may include clients and servers. Clients and servers are generally remote from each other and typically interact over a communications network. The relationship of client and server is created by computer programs running on corresponding computers and having a client-server relationship with each other.
  • the server can be a cloud server, also known as cloud computing server or cloud host. It is a host product in the cloud computing service system to solve the problem of traditional physical host and VPS service ("Virtual Private Server", or "VPS" for short) Among them, there are defects such as difficult management and weak business scalability.
  • the server can also be a distributed system server or a server combined with a blockchain.

Landscapes

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

Abstract

一种搜索结果的排序方法包括:获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序;在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式;基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分;按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。

Description

搜索结果的排序方法、装置、电子设备和存储介质
相关申请的交叉引用
本申请基于申请号为202210619839.8、申请日为2022年06月02日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本公开涉及计算机技术领域,具体涉及大数据等人工智能技术领域,尤其涉及一种搜索结果的排序方法、装置、电子设备和存储介质。
背景技术
随着互联网技术的不断发展,通过网络来搜索各类信息已经成为人们获取信息来源的主要方式之一。在搜索完成之后,搜索系统如何对搜索结果进行排序,从而既满足搜索需求,又实现信息推送,是目前亟需解决的问题。
发明内容
本公开旨在至少在一定程度上解决相关技术中的技术问题之一。
根据本公开第一方面,提出了一种搜索结果的排序方法,包括:
获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序;
在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式;
基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分;
按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
根据本公开第二方面,提出了一种搜索结果的排序装置,包括:
第一获取模块,用于获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序;
第一确定模块,用于在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式;
第二确定模块,用于基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分;
第二获取模块,用于按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
本公开第三方面实施例提出了一种计算机设备,包括:存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序,所述处理器执行所述程序时,实现如本公开第一方面实施例提出的方法。
本公开第四方面实施例提出了一种非临时性计算机可读存储介质,存储有计算机程序,所述计算机程序被处理器执行时实现如本公开第一方面实施例提出的方法。
本公开第五方面实施例提出了一种计算机程序产品,当所述计算机程序产品中的指令被处理器执行时,执行本公开第一方面实施例提出的方法。
本公开实施例中,首先获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序,之后在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式,然后基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分,之后按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。由此,在确定多个第一搜索结果中存在至少一个第一搜索结果为推送信息之后,可以对为推送信息的第一搜索结果进行再次评分,进而对各个第一搜索结果的初始排序进行更新和调整,使得最终的排序不仅能够满足用户的搜索需求,而且可以挖掘搜索结果的信息价值,实现信息推送,提高了排序结果所带来的价值。
本公开附加的方面和优点将在下面的描述中部分给出,部分将从下面的描述中变得明显,或通过本公开的实践了解到。
附图说明
本公开上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解,其中:
图1为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图;
图2为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图;
图3为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图;
图4为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图;
图5为本公开一些实施例提供的一种搜索结果的排序装置的结构示意图;
图6为用来实现本公开实施例的搜索结果的排序方法的电子设备的框图。
具体实施方式
下面详细描述本公开的实施例,所述实施例的示例在附图中示出,其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的,旨在用于解释本公开,而不能理解为对本公开的限制。
人工智能是研究使计算机来模拟人的某些思维过程和智能行为(如学习、推理、思考、规划等)的学科,既有硬件层面的技术也有软件层面的技术。人工智能硬件技术一般包括如传感器、专用人工智能芯片、云计算、分布式存储、大数据处理等技术;人工 智能软件技术主要包括计算机视觉技术、语音识别技术、自然语言处理技术以及机器学习、深度学习、大数据处理技术、知识图谱技术等几大方向。
大数据处理,是指采用人工智能的方式对规模巨大的数据进行分析以及处理的过程,而大数据可以概括为5个V,数据量大(Volume)、速度快(Velocity)、类型多(Variety)、价值(Value)、真实性(Veracity)。
下面结合参考附图描述本公开实施例的搜索结果的排序方法、装置、电子设备和存储介质。
本公开提供的一种搜索结果的排序方法,该方法可以由本公开提供的一种搜索结果的排序装置执行,也可以由本公开提供的电子设备执行,其中,电子设备可以包括但不限于台式电脑、平板电脑、云端设备、移动设备、个人数字助理等具有各种操作系统、触摸屏和/或显示屏的硬件设备,或者,也可以为服务器。下面以由本公开提供的搜索系统来执行本公开提供的一种搜索结果的排序方法,而不作为对本公开的限定。
图1为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图。
如图1所示,该搜索结果的排序方法,可以包括以下步骤:
步骤101,获取搜索请求对应的多个第一搜索结果及多个第一搜索结果的初始排序。
其中,搜索请求可以是搜索系统基于用户的输入的任一搜索文本而生成的,或者,也可以是基于用户点击的任一搜索项而生成的,在此不进行限定。
举例来说,若用户在输入框中输入“苹果”并触发回车键,则搜索系统则可以生成与“苹果”对应的搜索请求,并根据该搜索请求为用户展现对应的各个搜索结果。
其中,第一搜索结果可以为搜索系统(搜索引擎)基于用户的搜索请求查询所得到的搜索结果。本公开中,第一搜索结果可以为搜索结果页(Search engine results page,SERP),也即是指搜索系统对当前的搜索请求反馈的结果页面。
需要说明的是,在确定了搜索请求对应的各个第一搜索结果之后,搜索系统可以首先对各个第一搜索结果进行数据清洗,以过滤异常数据,之后可以基于各个第一搜索结果与搜索请求之间的相关程度,对各个第一搜索结果进行排序,进而生成初始排序。
本公开中,可以基于机器学习的算法,计算每个第一搜索结果与搜索请求之间的相关度,比如TF/IDF算法(Term Frequency&Inverse Document Frequency),在此不做限定。
举例来说,当前搜索请求Q对应的各个第一搜索结果分别为X、Y、Z,且X、Y、Z与搜索请求Q的相关度分别为0.05、0.6、0.3,则可以将Y-Z-X确定为初始排序,在此不做限定。
在一些实施例中,搜索系统还可以根据多个第一搜索结果中每个第一搜索结果与搜索请求的相关度,以及每个第一搜索结果在预设时段内的点击量,确定多个第一搜索结果的初始排序。
其中,预设时段可以为指定周期的时间段,比如可以为历史一个月,或者历史一个星期,在此不做限定。举例来说,若当前时间为4月1日,可以将3月1日到4月1日 之间的时间作为预设时段。
需要说明的是,不同时段的对某一搜索结果的点击量通常是不同的,也即是说,点击量通常具有时效性。因而,本公开中的预设时段可以设定为近期的历史时间段,比如距离当前时间一个星期内的时间。
举例来说,若预设时间段为最近一周,当前搜索请求Q对应的各个第一搜索结果分别为X、Y、Z,且X、Y、Z与搜索请求Q的相关度分别为0.8、0.6、0.2,而X、Y、Z在最近一周内的点击量为80、100、30,则进而可以计算X、Y、Z分别对应的得分,其中,X为0.8x80=64,Y为0.6x100=60,Z为0.2x30=6,则X、Y、Z对应的初始排序为X、Y、Z。其中,该举例仅为本公开的一种示意性说明,对本公开不构成限定。
由此,可以使得多个第一搜索结果的初始排序更加匹配用户的搜索意图,并且使得用户更方便地获取具有热门价值的内容。
步骤102,在至少一个第一搜索结果为推送信息的情况下,确定至少一个第一搜索结果的类型及样式。
其中,推送信息可以为具有商业价值的信息,比如搜索频率较高的信息,收入转化率较高的信息等等,在此不做限定。
其中,类型可以为综合页、列表页、内容页等等,在此不做限定。
其中,样式可以为图文、视频、卡片等等,在此不做限定。
作为一种可能实现的方式,可以将各个第一搜索结果输入至预先训练生成的神经网络模型中,以对各个第一搜索结果对应的价值进行标注,比如可以是“高价值”、“中价值”、“低价值”。进而将标注为“高价值”的第一搜索结果确定为推送信息。
需要说明的是,在确定了各个第一搜索结果中的推送信息之后,可以将推送信息对应的至少一个第一搜索结果进行排序。需要说明的是,在对至少一个第一搜索结果进行排序时,可以基于多个维度对第一搜索结果进行衡量,比如质量、时效性、相关性,在此不做限定。本公开中,可以根据该至少一个第一搜索结果的页面特征,对推送信息进行排序。
其中,页面特征可以包括类型以及样式等。
可以理解的是,不同类型及样式的搜索结果,用户的选择倾向通常是不同的,因而,本公开可以对当前推送信息对应的第一搜索结果的类型及样式进行判断,进而之后可以基于此进行评估。
步骤103,基于预设的映射关系,根据至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个第一搜索结果对应的第一得分。
其中,第一得分可以为对每种类型及样式的页面的评分。具体的,相同类型及样式的页面的第一得分是相同的,不同类型及样式的页面的第一得分可以相同,也可以不同。
需要说明的是,用户对于不同类型及样式的选择倾向通常是不同,比如一些用户在视频样式和文字样式的页面中,往往更倾向于选择视频样式的页面,因而对于视频样式的页面的点击率也比较高。在另一些情况中,一些用户在内容类型的页面和综合类型的 页面中,往往更倾向于选择内容类型的页面,本公开对此不做限定。因而,本公开可以根据历史数据中用户对各个类型及样式的页面的点击量或者收入转化率,预先设置每种类型和样式对应的得分,并通过该得分来表征用户对页面内容的喜爱程度以及价值高低。
在一些实施例中,可以预先设置一个映射关系表,并在该映射关系表中预设页面内容对应的类型及样式对应的第一得分。进而,搜索系统在确定当前的推送信息对应的各个第一搜索结果之后,可以根据各个第一搜索结果当前的类型和样式,确定每个第一搜索结果对应的第一得分。
步骤104,按照第一得分由大至小的顺序,对初始排序进行调整,以获取多个第一搜索结果更新后的排序。
可以理解的是,在至少一个第一搜索结果为推送信息之后,可以确定该至少一个第一搜索结果的第一得分,并根据第一得分由大到小的顺序对该至少一个第一搜索结果进行排序,并进一步地对初始排序进行调整。
举例来说,初始排序为r1,r2,r3,r4,r5,r6,r7。其中,r1,r3,r6为推送信息,且r1,r3,r6分别对应的第一得分为2,5,4,则可以将r1,r3,r6重新排序为r3,r6,r1,并对初始排序进行更新,进而获得更新后的排序为r3,r2,r6,r4,r5,r1,r7,在此不做限定。或者,还可以将推送信息对应的各个第一搜索结果放在最前面,也即可以获得更新后的排序为r3,r6,r1,r2,r4,r5,r7,在此不做限定。
本公开实施例中,首先获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序,之后在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式,然后基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分,之后按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。由此,在确定多个第一搜索结果中存在至少一个第一搜索结果为推送信息之后,可以对为推送信息的第一搜索结果进行再次评分,进而对各个第一搜索结果的初始排序进行更新和调整,使得最终的排序不仅能够满足用户的搜索需求,而且可以挖掘搜索结果的信息价值,实现信息推送,提高了排序结果所带来的价值。
图2为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图。
如图2所示,该搜索结果的排序方法,可以包括以下步骤:
步骤201,获取搜索请求对应的多个第一搜索结果。
需要说明的是,步骤201的具体实现方式可以参照上述实施例,在此不进行赘述。
步骤202,确定搜索请求当前映射的至少一个垂类标签。
其中,垂类标签可以为垂直领域分类标签。需要说明的是,一个搜索请求可能映射有一个或者多个不同的垂类标签,垂类标签可以用于表征当前搜索内容所属的领域或者行业。
举例来说,若当前搜索请求中包含的搜索文本为“百花”,则该搜索文本映射的垂类标签可以为{诗词}、{路名}、{花卉}、{奖项}、{影视},等等,在此不做限定。
或者,若当前搜索请求中包含的搜索文本为“新东方”,则该搜索文本映射的垂类标签可以为{烹饪学校}、{英语培训学校}、{方向}等等,在此不做限定。
步骤203,基于预设的价值指标,计算至少一个垂类标签中的每个垂类标签的第二得分。
其中,预设的价值指标可以有多个,比如点击率(CTR)、收入转化率(CVR)、搜索频率、平均点击价格(CPC)、平均转化价格(CPA)、每千次展现价格(CPM),在此不做限定。
其中,第二得分可以为对垂类标签的评分,用于表征垂类标签的价值高低,也即是说,若第二得分越高,对应的垂类标签的价值越大。
具体的,搜索系统可以根据预设的价值指标,获取与价值指标对应的各类数据,之后进行数据清洗,以过滤异常的数据,比如低质、作弊、空短、过期等数据。
本公开中,可以选取每个垂类标签对应的多个页面,并基于预设的价值指标,获取各个页面与价值指标对应的得分,进而根据每个页面在各项价值指标上的得分,以及每个价值指标对应的权重,计算每个垂类标签对应的得分。
具体的,可以利用以下公式进行计算:
value_score=sigmoid(∑w*value_info)
其中,value_score为垂类标签对应的第二得分,value_info为每个价值指标对应的得分,w为每个价值指标对应的权重,sigmoid为激活函数。
步骤204,确定每个第一搜索结果与每个垂类标签的相关度。
本公开中,可以基于机器学习的算法,计算每个第一搜索结果与每个垂类标签的相关度,比如TF/IDF算法(Term Frequency&Inverse Document Frequency),在此不做限定。
举例来说,若当前的第一搜索结果有3个,分别为a1,a2,a3,当前搜索请求映射的垂类标签有3个,分别为s1,s2,s3,则需要计算a1分别与s1,s2和s3之间的相关度,a2分别与s1,s2和s3之间的相关度,a3分别与s1,s2和s3之间的相关度。
需要说明的是,上述举例仅为一种示意性说明,对本公开不构成限定。
步骤205,根据每个第一搜索结果与每个垂类标签的相关度,及垂类标签的第二得分,确定每个第一搜索结果的第三得分。
其中,第三得分可以为第一搜索结果的初始排序得分。也即是说,本公开中,可以利用每个第一搜索结果的第三得分的大小,对各个第一搜索结果进行排序。
举例来说,若第一搜索结果有5个,分别为A、B、C、D、E,垂类标签有3个,分别为x、y、z,且x、y、z对应的垂类标签的第二得分分别为20、15、10。其中,A与x、y、z的相关度分别为0.8、0.7、0.6;B与x、y、z的相关度分别为0.9、0.2、0.2;C与x、y、z的相关度分别为0.6、0.6、0.6;D与x、y、z的相关度分别为0.2、0.8、0.1;E与x、y、z的相关度分别为0.5、0.8、0.8。
进一步地,则可以将每个第一搜索结果在每个垂类标签上的第二得分和相关度相乘, 并最后将各个相乘的结果相加,以计算每个第一搜索结果的第三得分的大小。
其中,A的第三得分为20x0.8+15x0.7+10x0.6=32.5,B的第三得分为20x0.9+15x0.2+10x0.2=23,C的第三得分为20x0.6+15x0.6+10x0.6=27,D的第三得分为20x0.2+15x0.8+10x0.1=17,E的第三得分为20x0.5+15x0.8+10x0.8=30。由此,可以根据A、B、C、D、E分别对应的第三得分32.5、23、27、17、30进行排序。
需要说明的是,上述举例仅为一种示意性说明,对本公开不构成限定。
步骤206,根据多个第一搜索结果分别对应的第三得分,确定多个第一搜索结果的初始排序。
举例来说,若当前各个第一搜索结果A、B、C、D、E对应的第三得分分别为32.5、23、27、17、30,则可以将A、B、C、D、E排为A、E、C、B、D。
需要说明的是,由此,可以使得初始排序既能考虑到第一搜索结果与当前搜索请求的相关性,还考虑到了搜索请求映射的各个垂类标签的价值,使得当前的初始排序能够满足用户搜索需求的同时,具有更高的商业价值。
步骤207,确定每个第一搜索结果对应的历史搜索频率以及历史收入转化率。
其中,历史搜索频率可以为指定历史时期内对任一搜索结果的搜索次数。
其中,收入转化率可以为指定历史时期内用户通过点击任一搜索结果,从而注册、激活或者成为付费用户的转化率。
需要说明的是,每个第一搜索结果在任一历史时间段内都具有一定的搜索频率以及收入转化率。本公开中,可以首先确定当前待统计的历史时间段,之后获取该时间段内第一搜索结果对应的历史搜索频率以及历史收入转化率。
步骤208,将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果,确定为推送信息。
其中,第一阈值可以为历史搜索频率的阈值,若历史搜索频率大于第一阈值,则说明该第一搜索结果比较热门,用户的搜索次数非常多,第一阈值的大小可以根据实际经验设定。
其中,第二阈值可以为历史收入转化率的阈值,若历史收入转化率大于第二阈值,则说明该第一搜索结果的转化率比较高,带来的收益比较大,第二阈值的大小可以根据实际经验设定。
需要说明的是,若任一历史搜索结果历史搜索频率较高或收入转化率较高,则说明该历史搜索结果比较受用户喜爱,为用户比较倾向选择的内容。
其中,推送信息可以为具有一定商业价值且用户更倾向选择的信息,通过向用户推送推送信息,可以使得用户能够接触到更有商业价值的内容。
在一些实施例中,可以将各个第一搜索结果中历史搜索频率大于第一阈值的第一搜索结果作为推送信息,或者,也可以将各个第一搜索结果中历史收入转化率大于第二阈值的第一搜索结果作为推送信息,或者,也可以将历史收入转化率大于第二阈值且历史搜索频率大于第一阈值的第一搜索结果作为推送信息,在此不做限定。
步骤209,在至少一个第一搜索结果为推送信息的情况下,确定至少一个第一搜索结果的类型及样式。
步骤210,基于预设的映射关系,根据至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个第一搜索结果对应的第一得分。
步骤211,按照第一得分由大至小的顺序,对初始排序进行调整,以获取多个第一搜索结果更新后的排序。
需要说明的是,步骤209、210、211的具体实现方式可以参照上述任一实施例,在此不进行赘述。
本公开实施例中,首先获取搜索请求对应的多个第一搜索结果,之后确定搜索请求当前映射的至少一个垂类标签,然后基于预设的价值指标,计算至少一个垂类标签中的每个垂类标签的第二得分,之后确定每个第一搜索结果与每个垂类标签的相关度,然后根据每个第一搜索结果与每个垂类标签的相关度,及垂类标签的第二得分,确定每个第一搜索结果的第三得分,之后根据多个第一搜索结果分别对应的第三得分,确定多个第一搜索结果的初始排序,然后确定每个所述第一搜索结果对应的历史搜索频率以及历史收入转化率,之后将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果,确定为推送信息,之后在至少一个第一搜索结果为推送信息的情况下,确定至少一个第一搜索结果的类型及样式,然后基于预设的映射关系,根据至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个第一搜索结果对应的第一得分,最后按照第一得分由大至小的顺序,对初始排序进行调整,以获取多个第一搜索结果更新后的排序。由此,通过将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果作为推送信息进行推送,从而可以将用户喜爱程度更高的页面优先展示。由于初始排序是根据每个第一搜索结果与垂类标签的相关度以及垂类标签的第二得分确定的,因而使得初始排序可以既能考虑到第一搜索结果与当前搜索请求的相关性,还考虑到了搜索请求映射的各个垂类标签的价值,使得当前的初始排序能够满足用户搜索需求的同时,具有更高的商业价值。
图3为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图。
如图3所示,该搜索结果的排序方法,可以包括以下步骤:
步骤301,获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序。
步骤302,在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式。
需要说明的是,步骤301、302的具体实现方式可以参照上述实施例,在此不进行赘述。
步骤303,获取指定历史时段的各个历史搜索结果。
其中,指定历史时段可以为预先选定的时间段,比如可以为距离当前日期一个星期以内的时间段,在此不做限定。需要说明的是,指定历史时段的终点可以为当前日期, 也即是说,指定历史时段是与当前日期相连的时间段,从而使得保障当前历史时段的各个历史搜索结果能够表征目前用户的选择倾向和个人意图,也即考虑了时效性对排序的影响。
其中,历史搜索结果可以为之前各个用户在搜索系统上搜索的内容。
步骤304,基于预设的价值指标,确定每个历史搜索结果对应的第四得分。
其中,预设的价值指标可以有多个,比如点击率(CTR)、收入转化率(CVR)、搜索频率、平均点击价格(CPC)、平均转化价格(CPA)、每千次展现价格(CPM),在此不做限定。
其中,第四得分可以为对历史搜索结果的评分,用于表征历史搜索结果的价值高低,也即是说,若第四得分越高,对应的历史搜索结果的价值越大。
具体的,搜索系统可以根据预设的价值指标,获取与价值指标对应的各类数据,之后进行数据清洗,以过滤异常的数据,比如低质、作弊、空短、过期等数据。
本公开中,搜索系统可以基于预设的价值指标,获取各个历史搜索结果与价值指标对应的得分,进而根据每个历史搜索结果在各项价值指标上的得分,以及每个价值指标对应的权重,计算每个历史搜索结果对应的得分。
具体的,可以利用以下公式进行计算:
value_score=sigmoid(∑w*value_info)
其中,value_score为历史搜索结果对应的第四得分,value_info为每个价值指标对应的得分,w为每个价值指标对应的权重,sigmoid为激活函数。
步骤305,根据具有相同类型及样式的各个历史搜索结果的第四得分均值,确定每种类型及样式与第一得分的映射关系。
需要说明的是,每个历史搜索结果都有对应的类型以及样式,其中,类型可以为内容页、列表页、综合页等等,在此不做限定。其中,样式可以为图文、视频、卡片等等,在此不做限定。
在获取了各个历史搜索结果之后,可以对各个历史搜索结果按照类型及样式进行分类,从而可以确定每类,也即每种相同类型及样式的历史搜索结果对应的第四得分的均值。
进一步地,可以将具有相同类型及样式的各个历史搜索结果的第四得分均值,确定为该种类型及样式对应的第一得分,也即是说,可以建立类型及样式和第一得分之间的映射关系。
步骤306,基于预设的映射关系,根据至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个第一搜索结果对应的第一得分。
步骤307,按照第一得分由大至小的顺序,对初始排序进行调整,以获取多个第一搜索结果更新后的排序。
需要说明的是,步骤306、307的具体实现方式可以参照上述实施例,在此不进行赘述。
步骤308,确定各个第一搜索结果当前的排列位置。
需要说明的是,在确定了各个第一搜索结果的排序之后,可以将每个第一搜索结果与当前排列位置对应起来。
举例来说,若各个第一搜索结果为Q1,Q2,Q3,Q4,Q5,Q6,Q7,则可以确定Q1,Q2,Q3,Q4,Q5,Q6,Q7分别对应的排列位置,比如将Q1,Q2,Q3,Q4分别排列在第一页的第一行,第一页的第二行,第一页的第三行,第一页的第四行,将Q5,Q6,Q7分别排列在第二页的第一行,第二页的第二行,第二页的第三行,在此不做限定。
步骤309,将第二搜索结果插入到所述排列位置中的指定展现位置中,其中,第二搜索结果为预先确定的具有指定页面特征或者指定页面内容的搜索结果。
其中,指定展现位置可以为预先指定的排列位置,比如首置位,或者第二置位,第三置位,在此不做限定。其中,指定展现位置可以为一个或者多个。
其中,第二搜索结果可以为具有指定页面特征或者指定页面内容的搜索结果,比如若根据当前搜索请求确定用户查询的搜索文本为“健康路”,则可以将“健康路”对应的百度地图作为第二搜索结果。或者,若根据当前搜索请求确定用户查询的搜索文本为“数学建模”,则可以将“数学建模”对应的百度百科作为第二搜索结果。或者,若根据当前搜索请求确定用户查询的搜索文本为“新闻联播”,则可以将视频页面特征的页面作为第二搜索结果,在此不进行限定。
在一些情况下,还可以基于搜索请求在目标名单库中进行查询,以确定是否有与所述搜索请求对应的合作方。其中,目标名单库中可以预设有各个合作名单。搜索系统若在目标名单库检索到与搜索请求对应的合作方,则可以根据合作方确定对应的页面内容,并将该页面内容确定为第二搜索结果。
在一些实施例中,在第二搜索结果为一个的情况下,可以将第二搜索结果插入到排列位置中的指定展现位置中。
举例来说,若指定展现位置为首置位,搜索系统则可以将第二搜索结果插入到原首置位的第一搜索结果的前面,并对当前的排序进行更新。
或者,在第二搜索结果为多个的情况下,可以将多个第二搜索结果按照预设的次序插入到排列位置中的各个指定展现位置中。
需要说明的是,在一些情况下,可以有多个指定展现位置。举例来说,若第二搜索结果有3个,则可以将该3个第二搜索结果按照预设的次序排列在第一置位、第二置位和第三置位,并对当前排序进行更新。
需要说明的是,上述举例仅为本公开的一种示意性说明,对本公开不构成限定。
本公开实施例中,首先获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序,之后在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式,然后获取指定历史时段的各个历史搜索结果,最后基于预设的价值指标,确定每个所述历史搜索结果对应的第四得分,之后根据具有相同类型及样式的各个所述历史搜索结果的第四得分均值,确定每种类型及样式与第一得分的映 射关系,然后基于预设的映射关系,根据至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个第一搜索结果对应的第一得分,之后按照第一得分由大至小的顺序,对初始排序进行调整,以获取多个第一搜索结果更新后的排序,然后确定各个所述第一搜索结果当前的排列位置,最后将第二搜索结果插入到所述排列位置中的指定展现位置中。由此,可以在对各个第一搜索结果进行排序之后,还可以将其他维度的第二搜索结果与第一搜索结果进行结合,从而可以提高排序结果的价值,更好的满足用户的搜索需求。
图4为本公开一些实施例提供的一种搜索结果的排序方法的流程示意图。
如图4所示,该搜索结果的排序方法,可以包括以下步骤:
步骤401,获取搜索请求对应的多个第一搜索结果及发起搜索请求的设备属性信息。
其中,搜索请求可以是搜索系统基于用户的输入的任一搜索文本而生成的,或者,也可以是基于用户点击的任一搜索项而生成的,在此不进行限定。
举例来说,若用户在输入框中输入“苹果”并触发回车键,则搜索系统则可以生成与“苹果”对应的搜索请求,并根据该搜索请求为用户展现对应的各个搜索结果。
其中,第一搜索结果可以为搜索系统(搜索引擎)基于用户的搜索请求查询所得到的搜索结果。本公开中,第一搜索结果可以为搜索结果页(Search engine results page,SERP),也即是指搜索系统对当前的搜索请求反馈的结果页面。
其中,发起搜索请求的设备属性信息可以为设备的相关信息或者设备使用者的相关信息,比如设备的类型,设备所使用的网络的类型,以及设备所属用户的信息,在此不进行限定。
在一些实施例中,设备属性信息包括以下至少一项:设备的类型,设备使用的网络类型,及设备所属的用户的属性信息。
其中,设备的类型可以为电脑、手机、平板等等,在此不做限定。
其中,设备使用的网络类型可以为局域网、城域网、广域网和互联网,在此不进行限定。
其中,用户的属性信息可以为用户的性别、年龄、职业、设备使用时间、页面操作记录信息,等等,在此不进行限定。
步骤402,根据预设的映射关系,确定与所述设备属性信息关联的参考排序。
其中,参考排序可以是对页面特征的排序,其中,页面特征可以包含页面类型和页面样式。
需要说明的是,发起搜索请求的设备在一定程度上会影响到用户浏览各个第一搜索结果的习惯,比如电脑设备相比与手机设备更适合播放视频类型的页面,平板电脑相比于手机设备更适合展现卡片类型的页面。
另外,年龄不同的设备使用用户对于页面类型的选择,也通常是不同的。女生喜欢浏览的页面类型跟男生喜欢浏览的页面类型之间也是不相同的。
因而,可以预先根据设备属性信息,确定对应的参考排序。比如,可以根据当前搜 索请求对应的设备的类型、设备使用的网络类型,及设备所属的用户的属性信息确定该搜索请求对应的唯一参考排序。
比如,可以通过根据预设的设备属性信息与参考排序之间的映射关系,确定当前设备属性信息对应的参考排序。其中,该映射关系可以是根据历史设备属性信息以及用户对搜索结果的操作记录确定的。
比如,对于设备属性信息A,其对应的参考排序可以为U、V、Z,对于设备属性信息B,其对应的参考排序可以为U、Z、V,对于设备属性信息C,其对应的参考排序可以为Z、V、U,在此不进行限定。其中,U、V、Z可以为不同的页面特征,比如长视频页面综合页面、卡片列表页面、短视频内容页,在此不进行限定。
步骤403,依据所述参考排序,确定所述多个第一搜索结果的初始排序。
在确定了参考排序之后,搜索系统可以根据各个第一搜索结果的页面特征,以及当前页面特征对应的参考排序,对各个第一搜索结果进行排序,从而可以确定多个第一搜索结果的初始排序。
步骤404,在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式。
步骤405,基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分。
步骤406,按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
需要说明的是,步骤404、405、406的具体实现方式可以参照上述实施例,在此不进行赘述。
本公开实施例中,首先获取所述搜索请求对应的多个第一搜索结果及发起所述搜索请求的设备属性信息,然后根据预设的映射关系,确定与所述设备属性信息关联的参考排序,之后依据所述参考排序,确定所述多个第一搜索结果的初始排序,然后在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式,之后基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分,最后按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。由此,可以根据搜索请求的设备属性信息以及预设的映射关系,确定参考排序,并进一步对第一搜索结果进行排序,从而确定一个更有参考价值的初始排序,之后再对各个第一搜索结果的初始排序进行更新和调整,使得最终的排序不仅能够满足用户的搜索需求,而且可以挖掘搜索结果的信息价值,实现信息推送,提高了排序结果所带来的价值。
图5是根据本公开一些实施例所提供的搜索结果的排序装置的结构示意图。
如图5所示,该搜索结果的排序装置50可以包括:第一获取模块510、第一确定模块520、第二确定模块530以及第二获取模块540。
第一获取模块510,用于获取搜索请求对应的多个第一搜索结果及所述多个第一搜 索结果的初始排序。
第一确定模块520,用于在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式。
第二确定模块530,用于基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分。
第二获取模块540,用于按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
在一些实施例中,所述第一获取模块510,具体用于:
根据所述多个第一搜索结果中每个第一搜索结果与所述搜索请求的相关度,以及每个所述第一搜索结果在预设时段内的点击量,确定所述多个第一搜索结果的初始排序。
在一些实施例中,所述第一获取模块510,具体用于:确定所述搜索请求当前映射的至少一个垂类标签;
基于预设的价值指标,计算所述至少一个垂类标签中的每个垂类标签的第二得分;
确定每个所述第一搜索结果与每个所述垂类标签的相关度;
根据每个所述第一搜索结果与每个所述垂类标签的相关度,及所述垂类标签的第二得分,确定每个所述第一搜索结果的第三得分;
根据所述多个第一搜索结果分别对应的第三得分,确定所述多个第一搜索结果的初始排序。
在一些实施例中,所述第一获取模块510,还用于:
确定每个所述第一搜索结果对应的历史搜索频率以及历史收入转化率;
将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果,确定为推送信息。
在一些实施例中,所述第二确定模块530,还用于:
获取指定历史时段的各个历史搜索结果;
基于预设的价值指标,确定每个所述历史搜索结果对应的第四得分;
根据具有相同类型及样式的各个所述历史搜索结果的第四得分均值,确定每种类型及样式与第一得分的映射关系。
在一些实施例中,所述第二获取模块540,包括:
确定单元,用于确定各个所述第一搜索结果当前的排列位置;
插入单元,用于将第二搜索结果插入到所述排列位置中的指定展现位置中,其中,所述第二搜索结果为预先确定的具有指定页面特征或者指定页面内容的搜索结果。
在一些实施例中,所述指定展现位置为一个或者多个,所述插入单元,具体用于:
在所述第二搜索结果为一个的情况下,将所述第二搜索结果插入到所述排列位置中的指定展现位置;
或者,
在所述第二搜索结果为多个的情况下,将多个所述第二搜索结果按照预设的次序插 入到所述排列位置中的各个所述指定展现位置中。
本公开实施例中,首先获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序,之后在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式,然后基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分,之后按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。由此,在确定多个第一搜索结果中存在至少一个第一搜索结果为推送信息之后,可以对为推送信息的第一搜索结果进行再次评分,进而对各个第一搜索结果的初始排序进行更新和调整,使得最终的排序不仅能够满足用户的搜索需求,而且可以挖掘搜索结果的信息价值,实现信息推送,提高了排序结果所带来的价值。
根据本公开的实施例,本公开还提供了一种电子设备、一种可读存储介质和一种计算机程序产品。
图6示出了可以用来实施本公开的实施例的示例电子设备500的示意性框图。电子设备旨在表示各种形式的数字计算机,诸如,膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算机。电子设备还可以表示各种形式的移动装置,诸如,个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例,并且不意在限制本文中描述的和/或者要求的本公开的实现。
如图6所示,设备500包括计算单元501,其可以根据存储在只读存储器(ROM)502中的计算机程序或者从存储单元508加载到随机访问存储器(RAM)503中的计算机程序,来执行各种适当的动作和处理。在RAM 503中,还可存储设备500操作所需的各种程序和数据。计算单元501、ROM 502以及RAM 503通过总线504彼此相连。输入/输出(I/O)接口505也连接至总线504。
设备500中的多个部件连接至I/O接口505,包括:输入单元506,例如键盘、鼠标等;输出单元507,例如各种类型的显示器、扬声器等;存储单元508,例如磁盘、光盘等;以及通信单元509,例如网卡、调制解调器、无线通信收发机等。通信单元509允许设备500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501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501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501执行上文所描述的各个方法和处理,例如图处理网络模型的训练方法。例如,在一些实施例中,图处理网络模型的训练方法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508。在一些实施例中,计算机程序的部分或者全部可以经由ROM 502和/或通信单元509而被载入和/或安装到设备500上。当计算机程序加载到RAM 503并 由计算单元501执行时,可以执行上文描述的图处理网络模型的训练方法的一个或多个步骤。备选地,在其他实施例中,计算单元501可以通过其他任何适当的方式(例如,借助于固件)而被配置为执行图处理网络模型的训练方法。
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、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)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括:实施在一个或者多个计算机程序中,该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释,该可编程处理器可以是专用或者通用可编程处理器,可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令,并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。
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。
为了提供与用户的交互,可以在计算机上实施此处描述的系统和技术,该计算机具有:用于向用户显示信息的显示装置(例如,CRT(阴极射线管)或者LCD(液晶显示器)监视器);以及键盘和指向装置(例如,鼠标或者轨迹球),用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互;例如,提供给用户的反馈可以是任何形式的传感反馈(例如,视觉反馈、听觉反馈、或者触觉反馈);并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如,作为数据服务器)、或者包括中间件部件的计算系统(例如,应用服务器)、或者包括前端部件的计算系统(例如,具有图形用户界面或者网络浏览器的用户计算机,用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括 这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如,通信网络)来将系统的部件相互连接。通信网络的示例包括:局域网(LAN)、广域网(WAN)、互联网和区块链网络。
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。服务器可以是云服务器,又称为云计算服务器或云主机,是云计算服务体系中的一项主机产品,以解决了传统物理主机与VPS服务("Virtual Private Server",或简称"VPS")中,存在的管理难度大,业务扩展性弱的缺陷。服务器也可以为分布式系统的服务器,或者是结合了区块链的服务器。
应该理解,可以使用上面所示的各种形式的流程,重新排序、增加或删除步骤。例如,本公开中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行,只要能够实现本公开公开的技术方案所期望的结果,本文在此不进行限制。
上述具体实施方式,并不构成对本公开保护范围的限制。本领域技术人员应该明白的是,根据设计要求和其他因素,可以进行各种修改、组合、子组合和替代。任何在本公开的精神和原则之内所作的修改、等同替换和改进等,均应包含在本公开保护范围之内。

Claims (21)

  1. 一种搜索结果的排序方法,包括:
    获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序;
    在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式;
    基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分;
    按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
  2. 根据权利要求1所述的方法,其中,所述获取搜索请求对应的所述多个第一搜索结果的初始排序,包括:
    根据所述多个第一搜索结果中每个第一搜索结果与所述搜索请求的相关度,以及每个所述第一搜索结果在预设时段内的点击量,确定所述多个第一搜索结果的初始排序。
  3. 根据权利要求1或2所述的方法,其中,所述获取搜索请求对应的所述多个第一搜索结果的初始排序,包括:
    确定所述搜索请求当前映射的至少一个垂类标签;
    基于预设的价值指标,计算所述至少一个垂类标签中的每个垂类标签的第二得分;
    确定每个所述第一搜索结果与每个所述垂类标签的相关度;
    根据每个所述第一搜索结果与每个所述垂类标签的相关度,及所述垂类标签的第二得分,确定每个所述第一搜索结果的第三得分;
    根据所述多个第一搜索结果分别对应的第三得分,确定所述多个第一搜索结果的初始排序。
  4. 根据权利要求1至3中任一项所述的方法,还包括:
    确定每个所述第一搜索结果对应的历史搜索频率以及历史收入转化率;
    将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果,确定为推送信息。
  5. 根据权利要求1至4中任一项所述的方法,还包括:
    获取指定历史时段的各个历史搜索结果;
    基于预设的价值指标,确定每个所述历史搜索结果对应的第四得分;
    根据具有相同类型及样式的各个所述历史搜索结果的第四得分均值,确定每种类型及样式与第一得分的映射关系。
  6. 根据权利要求1至5中任一项所述的方法,其中,在所述按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取多个所述第一搜索结果更新后的排序之后,还包括:
    确定各个所述第一搜索结果当前的排列位置;
    将第二搜索结果插入到所述排列位置中的指定展现位置中,其中,所述第二搜索结 果为预先确定的具有指定页面特征或者指定页面内容的搜索结果。
  7. 根据权利要求6所述的方法,其中,所述将第二搜索结果插入到所述排列位置中的指定展现位置中,包括:
    在所述第二搜索结果为一个的情况下,将所述第二搜索结果插入到所述排列位置中的指定展现位置;
    或者,
    在所述第二搜索结果为多个的情况下,将多个所述第二搜索结果按照预设的次序插入到所述排列位置中的各个所述指定展现位置中。
  8. 根据权利要求1-7中任一所述的方法,其中,所述获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序,包括:
    获取所述搜索请求对应的多个第一搜索结果及发起所述搜索请求的设备属性信息;
    根据预设的映射关系,确定与所述设备属性信息关联的参考排序;
    依据所述参考排序,确定所述多个第一搜索结果的初始排序。
  9. 根据权利要求8所述的方法,其中,所述设备属性信息包括以下至少一项:设备的类型,设备使用的网络类型,及设备所属的用户的属性信息。
  10. 一种搜索结果的排序装置,包括:
    第一获取模块,用于获取搜索请求对应的多个第一搜索结果及所述多个第一搜索结果的初始排序;
    第一确定模块,用于在至少一个第一搜索结果为推送信息的情况下,确定所述至少一个第一搜索结果的类型及样式;
    第二确定模块,用于基于预设的映射关系,根据所述至少一个第一搜索结果中每个第一搜索结果的类型及样式,确定每个所述第一搜索结果对应的第一得分;
    第二获取模块,用于按照所述第一得分由大至小的顺序,对所述初始排序进行调整,以获取所述多个第一搜索结果更新后的排序。
  11. 根据权利要求10所述的装置,其中,所述第一获取模块,具体用于:
    根据所述多个第一搜索结果中每个第一搜索结果与所述搜索请求的相关度,以及每个所述第一搜索结果在预设时段内的点击量,确定所述多个第一搜索结果的初始排序。
  12. 根据权利要求10或11所述的装置,其中,所述第一获取模块,具体用于:确定所述搜索请求当前映射的至少一个垂类标签;
    基于预设的价值指标,计算所述至少一个垂类标签中的每个垂类标签的第二得分;
    确定每个所述第一搜索结果与每个所述垂类标签的相关度;
    根据每个所述第一搜索结果与每个所述垂类标签的相关度,及所述垂类标签的第二得分,确定每个所述第一搜索结果的第三得分;
    根据所述多个第一搜索结果分别对应的第三得分,确定所述多个第一搜索结果的初始排序。
  13. 根据权利要求10至12中任一项所述的装置,其中,所述第一获取模块,还用 于:
    确定每个所述第一搜索结果对应的历史搜索频率以及历史收入转化率;
    将历史搜索频率大于第一阈值和/或历史收入转化率大于第二阈值的第一搜索结果,确定为推送信息。
  14. 根据权利要求10至13中任一项所述的装置,其中,所述第二确定模块,还用于:
    获取指定历史时段的各个历史搜索结果;
    基于预设的价值指标,确定每个所述历史搜索结果对应的第四得分;
    根据具有相同类型及样式的各个所述历史搜索结果的第四得分均值,确定每种类型及样式与第一得分的映射关系。
  15. 根据权利要求10至14中任一项所述的装置,其中,所述第二获取模块,包括:
    确定单元,用于确定各个所述第一搜索结果当前的排列位置;
    插入单元,用于将第二搜索结果插入到所述排列位置中的指定展现位置中,其中,所述第二搜索结果为预先确定的具有指定页面特征或者指定页面内容的搜索结果。
  16. 根据权利要求15所述的装置,其中,所述插入单元,具体用于:
    在所述第二搜索结果为一个的情况下,将所述第二搜索结果插入到所述排列位置中的指定展现位置;
    或者,
    在所述第二搜索结果为多个的情况下,将多个所述第二搜索结果按照预设的次序插入到所述排列位置中的各个所述指定展现位置中。
  17. 根据权利要求10-16中任一所述的装置,其中,所述第一获取模块,具体用于:
    获取所述搜索请求对应的多个第一搜索结果及发起所述搜索请求的设备属性信息;
    根据预设的映射关系,确定与所述设备属性信息关联的参考排序;
    依据所述参考排序,确定所述多个第一搜索结果的初始排序。
  18. 根据权利要求17所述的装置,其中,所述设备属性信息包括以下至少一项:设备的类型,设备使用的网络类型,及设备所属的用户的属性信息。
  19. 一种电子设备,包括:
    至少一个处理器;以及
    与所述至少一个处理器通信连接的存储器;其中,
    所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器能够执行权利要求1-9中任一项所述的方法。
  20. 一种存储有计算机指令的非瞬时计算机可读存储介质,其中,所述计算机指令用于使所述计算机执行根据权利要求1-9中任一项所述的方法。
  21. 一种计算机程序产品,包括计算机程序,所述计算机程序在被处理器执行时实现如权利要求1-9中任一项所述方法的步骤。
PCT/CN2022/128370 2022-06-02 2022-10-28 搜索结果的排序方法、装置、电子设备和存储介质 WO2023231288A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP22931264.0A EP4307136A1 (en) 2022-06-02 2022-10-28 Sorting method and apparatus for search results, and electronic device and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210619839.8 2022-06-02
CN202210619839.8A CN115017200B (zh) 2022-06-02 2022-06-02 搜索结果的排序方法、装置、电子设备和存储介质

Publications (1)

Publication Number Publication Date
WO2023231288A1 true WO2023231288A1 (zh) 2023-12-07

Family

ID=83072955

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/128370 WO2023231288A1 (zh) 2022-06-02 2022-10-28 搜索结果的排序方法、装置、电子设备和存储介质

Country Status (3)

Country Link
EP (1) EP4307136A1 (zh)
CN (1) CN115017200B (zh)
WO (1) WO2023231288A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115017200B (zh) * 2022-06-02 2023-08-25 北京百度网讯科技有限公司 搜索结果的排序方法、装置、电子设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130325838A1 (en) * 2012-06-05 2013-12-05 Yahoo! Inc. Method and system for presenting query results
CN112000871A (zh) * 2020-08-21 2020-11-27 北京三快在线科技有限公司 确定搜索结果列表的方法、装置、设备及存储介质
CN112925878A (zh) * 2019-12-06 2021-06-08 腾讯科技(深圳)有限公司 数据处理方法和装置
CN113934938A (zh) * 2021-10-22 2022-01-14 北京字跳网络技术有限公司 信息展示的方法、装置、可读介质和电子设备
CN115017200A (zh) * 2022-06-02 2022-09-06 北京百度网讯科技有限公司 搜索结果的排序方法、装置、电子设备和存储介质

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8583632B2 (en) * 2005-03-09 2013-11-12 Medio Systems, Inc. Method and system for active ranking of browser search engine results
US8489574B2 (en) * 2006-06-12 2013-07-16 Zalag Corporation Methods and apparatuses for searching content
CN101437039B (zh) * 2007-11-15 2012-11-07 华为技术有限公司 一种移动搜索的方法、系统和设备
CN103123653A (zh) * 2013-03-15 2013-05-29 山东浪潮齐鲁软件产业股份有限公司 基于贝叶斯分类学习的搜索引擎检索排序方法
US11017426B1 (en) * 2013-12-20 2021-05-25 BloomReach Inc. Content performance analytics
US11151642B2 (en) * 2019-01-02 2021-10-19 Michael Christian Smeltzer Method and system of electronic bartering
CN111581544A (zh) * 2019-04-08 2020-08-25 众巢医学科技(上海)股份有限公司 搜索结果优化方法、装置、计算机设备及存储介质
CN110688478B (zh) * 2019-09-29 2021-11-30 腾讯科技(深圳)有限公司 一种答案排序方法、装置及存储介质
CN111949898A (zh) * 2020-08-28 2020-11-17 平安国际智慧城市科技股份有限公司 搜索结果排序方法、装置、设备及计算机可读存储介质
CN113590917A (zh) * 2021-06-30 2021-11-02 五八有限公司 一种数据搜索方法、装置、电子设备及存储介质
CN114036355A (zh) * 2021-11-08 2022-02-11 北京库睿科技有限公司 一种搜索排序方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130325838A1 (en) * 2012-06-05 2013-12-05 Yahoo! Inc. Method and system for presenting query results
CN112925878A (zh) * 2019-12-06 2021-06-08 腾讯科技(深圳)有限公司 数据处理方法和装置
CN112000871A (zh) * 2020-08-21 2020-11-27 北京三快在线科技有限公司 确定搜索结果列表的方法、装置、设备及存储介质
CN113934938A (zh) * 2021-10-22 2022-01-14 北京字跳网络技术有限公司 信息展示的方法、装置、可读介质和电子设备
CN115017200A (zh) * 2022-06-02 2022-09-06 北京百度网讯科技有限公司 搜索结果的排序方法、装置、电子设备和存储介质

Also Published As

Publication number Publication date
CN115017200B (zh) 2023-08-25
EP4307136A1 (en) 2024-01-17
CN115017200A (zh) 2022-09-06

Similar Documents

Publication Publication Date Title
US10733197B2 (en) Method and apparatus for providing information based on artificial intelligence
RU2701110C2 (ru) Изучение и использование контекстных правил извлечения контента для устранения неоднозначности запросов
US11176142B2 (en) Method of data query based on evaluation and device
TWI567673B (zh) Sorting methods and devices for search results
US8725717B2 (en) System and method for identifying topics for short text communications
US10102191B2 (en) Propagation of changes in master content to variant content
US10713291B2 (en) Electronic document generation using data from disparate sources
US20120066073A1 (en) User interest analysis systems and methods
US10970293B2 (en) Ranking search result documents
US9367633B2 (en) Method or system for ranking related news predictions
CN109918555B (zh) 用于提供搜索建议的方法、装置、设备和介质
KR20090084853A (ko) 카테고리화를 통해 호스트 투 게스트 콘텐츠를 자동으로 매칭하기 위한 메커니즘
CN112733042A (zh) 推荐信息的生成方法、相关装置及计算机程序产品
US11977567B2 (en) Method of retrieving query, electronic device and medium
US11182540B2 (en) Passively suggesting text in an electronic document
WO2020123689A1 (en) Suggesting text in an electronic document
JP6593873B2 (ja) 情報分析装置及び情報分析方法
WO2015101161A1 (zh) 一种用于生成与目标系统对应的用户页面方法和装置
WO2023231288A1 (zh) 搜索结果的排序方法、装置、电子设备和存储介质
WO2023040230A1 (zh) 数据评估方法、训练方法、装置、电子设备以及存储介质
US20210271637A1 (en) Creating descriptors for business analytics applications
WO2023159766A1 (zh) 餐饮数据分析方法、装置、电子设备及存储介质
CN112839063A (zh) 消息推送方法、消息显示方法、存储介质和计算机设备
JP2019175212A (ja) 情報表示装置、記事ページ生成装置、情報処理装置、情報表示システム及びプログラム
CN114969542A (zh) 一种信息推送方法、服务端、装置和系统

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2022931264

Country of ref document: EP

Effective date: 20230922