Title core content determination method, search request processing method, and related devices

Abstract

The application provides a title core content determination method, a search request processing method and related devices, and relates to the field of artificial intelligence services. The title core content determination method comprises the following steps: first, obtaining a plurality of historical click records of a to-be-recommended album within a preset time range; then, based on the search text in each historical click record, obtaining the reference core content of the title of the to-be-recommended album and the type of the reference core content; if the reference core content is multiple and the type of each reference core content is different, determining the relationship between each reference core content and the remaining reference core content by using a pre-constructed knowledge graph; finally, according to the relationship between each reference core content and the remaining reference core content, obtaining the reference core content that meets the preset condition and taking it as the core content of the title of the to-be-recommended album, thereby avoiding the interference of the content that rides on the popularity in the title of the album on the search engine and enabling the search engine to accurately understand the core content of the title.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence services, in particular to a title core content determination method, a search request processing method and related devices. BACKGROUND

[0002] In a content-based (e.g., books, videos, audio) search scenario, the title of each album contains key information, for example, in an audio-based search scenario, the title of each audio generally contains the title of the work, the author, the host, and other information. When the user inputs specific query content, the search engine generally determines the album that best matches the user's intention based on the title of each album to push to the user.

[0003] Due to different title expression methods, the push order of recommended content is affected, for example, some authors or hosts want their recommended content to be pushed to the user in a position in the front, and may add some keywords of other popular hosts or works in the title, but this will interfere with the search engine's understanding of the core content of the title, so that the recommended content pushed to the user in the front does not match the user's intention, affecting the user experience. SUMMARY

[0004] In order to overcome the shortcomings of the prior art, the embodiments of the present application provide a title core content determination method, a search request processing method and related devices.

[0005] The technical solution of the embodiments of the present application can be implemented as follows:

[0006] In a first aspect, the embodiments of the present application provide a title core content determination method, which comprises:

[0007] Obtaining a plurality of historical click records of a to-be-recommended album within a preset time range, wherein each historical click record includes search text input by a user;

[0008] Based on the search text in each historical click record, obtaining reference core content of the title of the to-be-recommended album and the type of the reference core content;

[0009] If the reference core content is multiple and the type of each reference core content is different, then using a pre-constructed knowledge graph to determine the relationship between each reference core content and the remaining reference core content;

[0010] According to the relationship between each reference core content and the remaining reference core content, obtaining reference core content that meets a preset condition, and taking the reference core content that meets the preset condition as the core content of the title of the to-be-recommended album.

[0011] Optionally, the step of obtaining the reference core content of the title of the album to be recommended and the type of the reference core content based on the search text in each of the historical click records comprises:

[0012] obtaining each pending core content of the title of the album to be recommended according to the title of the album to be recommended and the search text in each of the historical click records;

[0013] processing the search text in each of the historical click records by using a pre-constructed dictionary and a pre-trained entity extraction model to obtain at least one entity set;

[0014] obtaining the reference core content and the type of the reference core content based on each of the pending core content and each of the entity sets.

[0015] Optionally, the step of obtaining at least one pending core content of the title of the album to be recommended according to the title of the album to be recommended and the search text in each of the historical click records comprises:

[0016] performing deduplication processing on the search text in the plurality of historical click records to obtain at least one candidate text;

[0017] for each of the candidate texts, taking the ratio of the number of historical click records in which the search text is the candidate text to the total number of historical click records as the album click rate corresponding to the candidate text to obtain the album click rate corresponding to each of the candidate texts;

[0018] determining each target text from all candidate texts according to the album click rate;

[0019] taking the overlapping part of each of the target texts and the title of the album to be recommended as one of the pending core contents.

[0020] Optionally, each of the entity sets has a corresponding entity type, and the step of obtaining the reference core content and the type of the reference core content based on each of the pending core content and each of the entity sets comprises:

[0021] for each of the pending core contents, if a target set exists in all entity sets, taking the pending core content as one of the reference core contents and taking the entity type corresponding to the target set as the type of the reference core content, wherein the target set has an entity with the same name as the pending core content;

[0022] if the target set does not exist in all entity sets, retaining the pending core content;

[0023] If none of the core contents can be the reference core content, the title of the album to be recommended is taken as the reference core content, and a type of the reference core content is set as a preset type.

[0024] Optionally, the step of obtaining the reference core content satisfying the preset condition according to the relationship between each reference core content and the rest of the reference core contents comprises:

[0025] For each reference core content, if there is the rest of the reference core contents related to the reference core content, it is determined that the reference core content satisfies the preset condition.

[0026] If there is no rest of the reference core contents related to the reference core content, it is determined that the reference core content does not satisfy the preset condition.

[0027] In a second aspect, an embodiment of the present application provides a search request processing method, and the method comprises:

[0028] Obtaining a user search request, wherein the user search request comprises search text input by a user;

[0029] Calculating similarity between the search text and core content of a title of each album to be pushed, wherein the core content of the title of the album to be pushed is obtained by using the title core content determination method in the first aspect;

[0030] According to the size of the similarity, each album to be pushed is pushed in sequence to respond to the user search request.

[0031] In a third aspect, an embodiment of the present application provides a title core content determination apparatus, and the apparatus comprises:

[0032] A first obtaining module is configured to obtain a plurality of historical click records of an album to be recommended within a preset time range, wherein each historical click record comprises search text input by a user;

[0033] A first processing module is configured to:

[0034] Based on the search text in each historical click record, reference core content of the title of the album to be recommended and a type of the reference core content are obtained;

[0035] If the reference core content is multiple and the type of each reference core content is different, a relationship between each reference core content and the rest of the reference core contents is determined by using a pre-constructed knowledge graph;

[0036] Based on the relationship between each of the reference core contents and the rest of the reference core contents, reference core contents that meet preset conditions are obtained, and the reference core contents that meet the preset conditions are used as the core contents of the title of the album to be recommended.

[0037] Fourthly, embodiments of the present invention provide a search request processing apparatus, characterized in that the apparatus comprises:

[0038] The second acquisition module is used to acquire user search requests, which include search text input by the user.

[0039] The second processing module is used to calculate the similarity between the search text and the core content of the title of each album to be pushed, wherein the core content of the title of the album to be pushed is obtained using the title core content determination method as described in the first aspect.

[0040] The push module is used to push each of the albums to be pushed in sequence according to the similarity, in response to the user's search request.

[0041] Fifthly, embodiments of the present invention provide a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the title core content determination method as described in the first aspect, and / or the search request processing method as described in the second aspect.

[0042] In a sixth aspect, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the title core content determination method as described in the first aspect, and / or the search request processing method as described in the second aspect.

[0043] Compared with the prior art, the title core content determination method provided by the embodiment of the application firstly acquires a plurality of historical click records of the album to be recommended within a preset time range, wherein each historical click record includes search text input by a user; then, based on the search text in each historical click record, reference core content of the title of the album to be recommended and the type of the reference core content are obtained; if the reference core content is multiple and the type of each reference core content is different, the relationship between each reference core content and the remaining reference core content is determined by using a pre-constructed knowledge graph; finally, according to the relationship between each reference core content and the remaining reference core content, reference core content satisfying a preset condition is obtained, and the reference core content satisfying the preset condition is taken as the core content of the title of the album to be recommended. Since the embodiment of the application determines the relationship between each reference core content and the remaining reference core content by using the pre-constructed knowledge graph when the determined reference core content is multiple and the type of each reference core content is different, and then obtains the reference core content satisfying the preset condition from the relationship between each reference core content and the remaining reference core content to be taken as the core content of the title of the album to be recommended, the interference of the content of the title of the album to search engine is avoided, so that the search engine can accurately understand the core content of the title. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0045] Figure 1 The flowchart of the title core content determination method provided by the embodiment of the application;

[0046] Figure 2 The flowchart of the implementation of step S102 provided by the embodiment of the application;

[0047] Figure 3 The flowchart of the search request processing method provided by the embodiment of the application;

[0048] Figure 4 The function unit block diagram of the title core content determination device provided by the embodiment of the application;

[0049] Figure 5 The function unit block diagram of the search request processing device provided by the embodiment of the application;

[0050] Figure 6A structural schematic block diagram of a computer device is provided for an embodiment of the present application.

[0051] Icon: 100 - title core content determination apparatus; 101 - first acquisition module; 102 - first processing module; 200 - search request processing apparatus; 201 - second acquisition module; 202 - second processing module; 203 - pushing module; 300 - computer device; 310 - memory; 320 - processor. DETAILED DESCRIPTION

[0052] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0054] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, thus, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0055] It should be noted that: the features in the embodiments of the present application can be combined with each other without conflict.

[0056] In order to avoid the interference of the hot content in the title of the album to the search engine, so that the search engine can accurately understand the core content of the title, the title core content determination method provided in the embodiments of the present application will be described in detail below.

[0057] Please refer to Figure 1 The title core content determination method includes steps S101-S104.

[0058] S101, a plurality of historical click records of the album to be recommended in a preset time range are acquired.

[0059] The preset time range can be one day, one week or one month, and can be set according to actual needs. Each historical click record includes a search text input by a user.

[0060] Understandably, when a user enters search text through the client, the backend search engine pushes multiple albums to the user based on the search text. When the user clicks on one of the albums, the backend will generate a historical click record for the album that the user clicked. This historical click record includes the user's click time, the search text entered by the user, and other information.

[0061] For example, suppose a user enters the search text "Mi Xiaoquan" through the client. The backend search engine pushes an album to the user based on the search text "Mi Xiaoquan", which includes "Xiao Dabao's Fun Reading of 'Mi Xiaoquan's School Diary'". The user clicks on the album. At this time, the backend generates a historical click record for the album "Xiao Dabao's Fun Reading of 'Mi Xiaoquan's School Diary'", including the user's click time and the search text "Mi Xiaoquan".

[0062] S102, based on the search text in each historical click record, obtain the reference core content of the title of the album to be recommended and the type of the reference core content.

[0063] The reference core content is the overlapping part of the search text and the title of the album to be recommended. The type of reference core content can be author, work, broadcaster, or character, etc.

[0064] For example, suppose the search text in the historical click record is "Xiao Dabao", the title of the album to be recommended is "Xiao Dabao's Interesting Reading of 'Mi Xiaoquan's School Diary'", and the core reference content obtained is "Xiao Dabao", the type of which is broadcaster.

[0065] S103 If there are multiple reference core contents and each reference core content is of a different type, then the relationship between each reference core content and the other reference core contents is determined by using a pre-constructed knowledge graph.

[0066] Among them, the knowledge graph is a knowledge graph that constructs relationships between authors, works, roles, institutions, styles, etc. within the site's content, and it can be pre-built.

[0067] The construction process of a knowledge graph includes knowledge extraction, knowledge fusion, knowledge processing, and knowledge updating. In this embodiment of the invention, no specific limitation is made on the construction method of the knowledge graph.

[0068] For album titles that may contain content that piggybacks on trending topics, such as keywords from other popular streamers or works, knowledge graphs can be used to identify and remove these elements when they are extracted as core content for the album titles, thus preventing them from being used as the core content of the album titles.

[0069] Exemplarily, it is assumed that the title of the album to be recommended is "Mountain Village Ghost Story Purple Cuff", wherein "Mountain Village Ghost Story" has no relationship with "Purple Cuff", because "Purple Cuff" is a very famous host, and therefore the author of the album to be recommended adds "Purple Cuff" into the title in order to ride on the popularity of the popular host.

[0070] For the album to be recommended, if the two extracted reference core contents are "Mountain Village Ghost Story" and "Purple Cuff", respectively, and the types are works and hosts, respectively, at this time, it is queried whether the author and the host of "Mountain Village Ghost Story" contain "Purple Cuff", or it is also queried whether "Purple Cuff" contains the work "Mountain Village Ghost Story", and if the query result is that "Purple Cuff" does not contain "Mountain Village Ghost Story", then "Purple Cuff" is deleted.

[0071] In S104, the reference core content satisfying the preset condition is obtained according to the relationship between each reference core content and the remaining reference core contents, and the reference core content satisfying the preset condition is taken as the core content of the title of the album to be recommended.

[0072] The relationship between each reference core content and the remaining reference core contents includes correlation and irrelevance. The preset condition means that for any reference core content, there is a remaining reference core content related to it.

[0073] In the embodiment of the application, the implementation process of "obtaining the reference core content satisfying the preset condition according to the relationship between each reference core content and the remaining reference core contents" in S104 can be as follows:

[0074] For each reference core content, if there is a remaining reference core content related to the reference core content, it is determined that the reference core content satisfies the preset condition.

[0075] If there is no remaining reference core content related to the reference core content, it is determined that the reference core content does not satisfy the preset condition.

[0076] Exemplarily, it is assumed that the three reference core contents of the album to be recommended are "Xiao Dabao" and "Rice Small Circle School Record", and the types are hosts and works, respectively. By querying "Xiao Dabao" in the knowledge graph, it is found that it contains the work "Rice Small Circle School Record", and by querying "Rice Small Circle School Record" in the knowledge graph, it is found that the host is "Xiao Dabao".

[0077] The reference core contents "Xiao Dabao" and "Rice Small Circle School Record" are related, and both satisfy the preset condition, and therefore the core content of the title of the album to be recommended is "Xiao Dabao" and "Rice Small Circle School Record".

[0078] The step S102 will be described in detail below.

[0079] Please refer to Figure 2Step S102 includes sub-steps S102-1-S102-3.

[0080] S102-1, according to the title of the album to be recommended and the search text in each historical click record, obtaining each pending core content of the title of the album to be recommended.

[0081] Optionally, the implementation process of step S102-1 can include sub-steps S102-1-1-S102-1-4.

[0082] S102-1-1, performing deduplication processing on the search text in the plurality of historical click records to obtain at least one candidate text.

[0083] The candidate text is the result of deduplication processing on the search text in the plurality of historical click records.

[0084] For example, it is assumed that within a preset time range, the album to be recommended with the title "Xiaodabao interesting reading of 'Miaocircle's school record'" has a total of 5 historical click records, wherein the search text in the 1st and 3rd historical click records is "Miaocircle", the search text in the 2nd and 4th historical click records is "Xiaodabao", and the search text in the 5th historical click record is "Miaocircle's school record".

[0085] Deduplication processing is performed on the search text in the 5 historical click records, and "Miaocircle", "Xiaodabao", and "Miaocircle's school record" are taken as candidate texts.

[0086] S102-1-2, for each candidate text, taking the ratio of the number of historical click records with search text as the candidate text to the total number of historical click records as the album click rate corresponding to the candidate text, and obtaining the album click rate corresponding to each candidate text.

[0087] For example, it is assumed that the album to be recommended with the title "Xiaodabao interesting reading of 'Miaocircle's school record'" has a total of 5 historical click records, wherein the search text in the 1st and 3rd historical click records is "Miaocircle", the search text in the 2nd and 4th historical click records is "Xiaodabao", and the search text in the 5th historical click record is "Miaocircle's school record", and through step S102-1-1, three candidate texts can be determined, which are "Miaocircle", "Xiaodabao", and "Miaocircle's school record".

[0088] Since the search text is the historical click record of the candidate text "MIXIAOQUAN", there are two, so the album click rate corresponding to the candidate text "MIXIAOQUAN" is (2 / 5), that is, 0.4. Similarly, since the search text is the historical click record of the candidate text "XIAODABAO", there are two, so the album click rate corresponding to the candidate text "XIAODABAO" is (2 / 5), that is, 0.4. Since the search text is the historical click record of the candidate text "MIXIAOQUANXUESHIJI", there is one, so the album click rate corresponding to the candidate text "MIXIAOQUANXUESHIJI" is (1 / 5), that is, 0.2.

[0089] S102-1-3, determining each target text from all candidate texts according to the album click rate.

[0090] Among them, the target text refers to the candidate text with an album click rate greater than a preset threshold, and the preset threshold can be set according to requirements. In a possible implementation manner, the preset threshold can be set to 0.01, which means that the candidate text with an album click rate greater than 0.01 can be used as a target text.

[0091] S102-1-4, taking each target text and the overlapping part of the title of the album to be recommended as a said pending core content.

[0092] Exemplarily, assuming that the title of the album to be recommended is "XIAODABAOQUQIDU 'MIXIAOQUANXUESHIJI'", the target texts determined by the search texts in the historical click records include "XIAODABAO" and "MIXIAOQUANXUESHIJI", and the target texts "XIAODABAO" and "MIXIAOQUANXUESHIJI" are respectively taken as the overlapping part of the title of the album to be recommended "XIAODABAOQUQIDU 'MIXIAOQUANXUESHIJI'", to obtain two pending core contents, which are "XIAODABAO" and "MIXIAOQUANXUESHIJI" respectively.

[0093] S102-2, processing the search text in each historical click record by using a pre-constructed dictionary and a pre-trained entity extraction model to obtain at least one entity set.

[0094] In the embodiment of the present application, the author, work, anchor, main character and other entities can be mined through an off-site knowledge graph (for example, ownthink), and then added to the dictionary after cleaning to realize the construction of the dictionary. The album title and album description in the station can be used as training corpus, and after artificial annotation, the bert model is trained to obtain the entity extraction model.

[0095] For each search text input by the user, the entities are extracted from the search text by using the pre-constructed dictionary and the pre-trained entity extraction model respectively, and the entities extracted by using the dictionary and the entities extracted by using the model are fused and de-duplicated, and all the entities after fusion and de-duplication are divided into different entity sets according to the entity type.

[0096] Exemplarily, assuming that the search text input by the user is "Three Body Liu Cixin", the dictionary and the model both extract the entity "Liu Cixin", the results extracted by the dictionary and the model are fused and de-duplicated to obtain the entity "Liu Cixin", and since the entity type of the entity "Liu Cixin" is author, the entity "Liu Cixin" is divided into the entity set corresponding to the entity type of author.

[0097] S102-3, based on each pending core content and each entity set, obtaining a reference core content and a type of the reference core content.

[0098] Optionally, the implementation process of step S102-3 can include sub-steps S102-3-1-S102-3-3.

[0099] S102-3-1, for each pending core content, if there is a target set in all entity sets, the pending core content is taken as a reference core content, and the entity type corresponding to the target set is taken as the type of the reference core content.

[0100] Wherein, each entity set has a corresponding entity type, and the target set has an entity with the same name as the pending core content.

[0101] Exemplarily, assuming that the pending core content of the album to be recommended with the title "Xiaodabao interesting reading of 'Miaoxiaojianjijil'" is "Xiaodabao" and "Miaoxiaojianjijil", the entity set 1 and the entity set 2 are obtained from the search text in the historical click record of the album through step S102-2, wherein the entity type corresponding to the entity set 1 is works, and the entity type corresponding to the entity set 2 is anchor.

[0102] Since there is an entity with the name "Xiaodabao" in the entity set 2, and there is an entity with the name "Miaoxiaojianjijil" in the entity set 1, for the pending core content "Xiaodabao", there is a target set, that is, the entity set 2, and for the pending core content "Miaoxiaojianjijil", there is a target set, that is, the entity set 1, so two reference core contents can be obtained, which are "Xiaodabao" with the type of anchor and "Miaoxiaojianjijil" with the type of works, respectively.

[0103] S102-3-2, if there is no target set in all entity sets, the pending core content is retained.

[0104] It can be understood that for any pending core content, if there is no entity with the name of the pending core content in each entity set, it means that the pending core content cannot be taken as a reference core content.

[0105] S102-3-3, if each core content cannot be used as a reference core content, the title of the album to be recommended is used as the reference core content, and the type of the reference core content is set to a preset type.

[0106] For example, assuming that the pending core content of the album to be recommended with the title of "Three-body author recommendation | Tomorrow warship" is "Three-body" and "Tomorrow warship", and a set of entities of a type of work is obtained from the search text in the historical click record of the album by step S102-2.

[0107] Since there is no entity with the entity name of "Three-body" and no entity with the entity name of "Tomorrow warship" in the set of entities, the pending core content of "Three-body" and "Tomorrow warship" cannot be used as the reference core content, at this time, the title "Three-body author recommendation | Tomorrow warship" is used as the reference core content of the album to be recommended as a whole, and the type of the reference core content is set to a preset type.

[0108] The embodiment of the application further provides a search request processing method, which will be described in detail below.

[0109] Please refer to Figure 3 The search request processing method comprises steps S201-S203.

[0110] S201, obtaining a user search request.

[0111] The user search request comprises search text input by a user.

[0112] S202, calculating the similarity between the search text and the core content of the title of each album to be pushed.

[0113] The core content of the title of the album to be pushed is obtained by using the title core content determination method.

[0114] It should be noted that for the album to be pushed with more than one core content, the weight of each core content in the calculation of the similarity is determined according to the album click rate of the candidate text used to obtain each core content in step S102-1-2.

[0115] In a possible implementation manner, the album click rate of the candidate text corresponding to each core content can be normalized to obtain the weight of each core content.

[0116] Exemplarily, assuming that the album to be recommended with the title of "Xiaodabao's interesting reading of 'Miaoxiaohuan's school record'" has the core content of "Xiaodabao" and "Miaoxiaohuan's school record", and the album click rates of the corresponding candidate texts are 0.311 and 0.444 respectively, by performing normalization processing on the album click rates, it is obtained that the weight of the core content "Xiaodabao" is 0.311 / (0.444+0.311)=0.412 and the weight of the core content "Miaoxiaohuan's school record" is 0.444 / (0.444+0.311)=0.588 when calculating the similarity.

[0117] S203, in response to the user search request, sequentially pushing each of the to-be-pushed album according to the size of the similarity.

[0118] Compared with the prior art, the embodiment of the present application has the following technical effects:

[0119] (1) In the case where the determined reference core content is multiple and the type of each reference core content is different, the relationship between each reference core content and the remaining reference core content is determined by using the pre-constructed knowledge graph, and then the reference core content satisfying the preset condition is obtained from the relationship between each reference core content and the remaining reference core content, to serve as the core content of the title of the album to be recommended, thereby avoiding the interference of the hot content in the title of the album to the search engine, so that the search engine can accurately understand the core content of the title.

[0120] (2) The push order of each to-be-pushed album is determined by using the similarity between the search text and the core content of the title of each to-be-pushed album, so that the album that matches the user's intention is pushed in the front, thereby improving the user experience.

[0121] In order to perform the corresponding steps in the above-mentioned method embodiments and various possible implementations, the following respectively gives an implementation of a title core content determination device 100 and a search request processing device 200.

[0122] Please refer to Figure 4 The title core content determination device 100 comprises a first acquisition module 101 and a first processing module 102.

[0123] The first acquisition module 101 is configured to acquire a plurality of historical click records of the album to be recommended within a preset time range, wherein each historical click record comprises a search text input by a user.

[0124] The first processing module 102 is configured to obtain a reference core content of a title of the album to be recommended and a type of the reference core content based on search text in each historical click record; if the reference core content is multiple and the type of each reference core content is different, determine a relationship between each reference core content and the remaining reference core content by using a pre-constructed knowledge graph; and obtain a reference core content satisfying a preset condition according to the relationship between each reference core content and the remaining reference core content, and take the reference core content satisfying the preset condition as the core content of the title of the album to be recommended.

[0125] Optionally, the first processing module 102 is specifically configured to obtain each pending core content of the title of the album to be recommended according to the title of the album to be recommended and the search text in each historical click record; process the search text in each historical click record by using a pre-constructed dictionary and a pre-trained entity extraction model to obtain at least one entity set; and obtain the reference core content and the type of the reference core content based on each pending core content and each entity set.

[0126] Optionally, when the first processing module 102 is configured to obtain at least one pending core content of the title of the album to be recommended according to the title of the album to be recommended and the search text in each historical click record, the first processing module 102 is specifically configured to perform deduplication processing on the search text in the multiple historical click records to obtain at least one candidate text; for each candidate text, take a ratio of a number of historical click records with the search text being the candidate text to a total number of historical click records as an album click rate corresponding to the candidate text to obtain the album click rate corresponding to each candidate text; determine each target text from all candidate texts according to the album click rate; and take an overlapping part of each target text and the title of the album to be recommended as one of the pending core contents.

[0127] Optionally, each entity set has a corresponding entity type, and when the first processing module 102 is configured to obtain the reference core content and the type of the reference core content based on each pending core content and each entity set, the first processing module 102 is specifically configured to, for each pending core content, if there is a target set in all entity sets, take the pending core content as a reference core content and take an entity type corresponding to the target set as the type of the reference core content, wherein the target set has an entity with the same name as the pending core content; if there is no target set in all entity sets, keep the pending core content; and if each core content cannot be taken as a reference core content, take the title of the album to be recommended as the reference core content and set the type of the reference core content as a preset type.

[0128] Optionally, the first processing module 102 is further specifically configured to determine, for each reference core content, that the reference core content meets the preset condition if there is remaining reference core content related to the reference core content; and determine that the reference core content does not meet the preset condition if there is no remaining reference core content related to the reference core content.

[0129] Please refer to Figure 5 , the search request processing apparatus 200 includes a second acquisition module 201, a second processing module 202 and a pushing module 203.

[0130] The second acquisition module 201 is configured to acquire a user search request, the user search request including a search text input by a user.

[0131] The second processing module 202 is configured to calculate a similarity between the search text and a core content of a title of each album to be pushed, wherein the core content of the title of the album to be pushed is obtained by using the title core content determination method as described above.

[0132] The pushing module 203 is configured to push each album to be pushed in sequence according to the size of the similarity to respond to the user search request.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the title core content determination apparatus 100 and the search request processing apparatus 200 described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0134] Further, the embodiments of the present application also provide a computer device 300, please refer to Figure 6 The computer device 300 can include a memory 310 and a processor 320.

[0135] The processor 320 can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs of the title core content determination method and / or the search request processing method provided in the foregoing method embodiments.

[0136] The memory 310 can be a ROM, or other type of static storage devices that can store static information and instructions, a RAM, or other type of dynamic storage device that can store information and instructions, an Electrically Erasable Programmable-Only Memory (EEPROM), a Compact disc Read-Only Memory (CD-ROM) or other optical disk storage, a magnetic disk storage or other magnetic storage devices, or any other medium capable of storing desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto. The memory 310 can exist independently, and is connected to the processor 320 through a communication bus. The memory 310 can also be integrated with the processor 320. The memory 310 is configured to store machine executable instructions for implementing the solutions of the present application. The processor 320 is configured to execute the machine executable instructions stored in the memory 310, so as to implement the above-mentioned method embodiments.

[0137] The present application also provides a computer readable storage medium containing a computer program, which, when executed, can be used to perform the title core content determination method provided by the above-mentioned method embodiments, and / or the related operations in the search request processing method.

[0138] To sum up, the title core content determination method provided by the embodiments of the present application first acquires a plurality of historical click records of the album to be recommended within a preset time range, wherein each historical click record includes a search text input by a user; then, based on the search text in each historical click record, the reference core content of the title of the album to be recommended and the type of the reference core content are obtained; if the reference core content is multiple and the type of each reference core content is different, the relationship between each reference core content and the remaining reference core content is determined by using the pre-constructed knowledge graph; finally, the reference core content satisfying the preset condition is obtained according to the relationship between each reference core content and the remaining reference core content, and the reference core content satisfying the preset condition is taken as the core content of the title of the album to be recommended. Since the embodiments of the present application determine the relationship between each reference core content and the remaining reference core content by using the pre-constructed knowledge graph when the determined reference core content is multiple and the type of each reference core content is different, and then obtain the reference core content satisfying the preset condition from the relationship between each reference core content and the remaining reference core content to be used as the core content of the title of the album to be recommended, the content of the title of the album to be recommended that is used to ride on the popularity of others is avoided to interfere with the search engine, so that the search engine can accurately understand the core content of the title.

[0139] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A title core content determination method characterized by comprising: The method comprises: obtaining a plurality of historical click records of a to-be-recommended album within a preset time range, wherein each historical click record comprises search text input by a user; based on the search text in each historical click record, obtaining a reference core content of the title of the to-be-recommended album and a type of the reference core content; if the reference core content is multiple and the type of each reference core content is different, determining the relationship between each reference core content and the remaining reference core content by using a pre-constructed knowledge graph; for each reference core content, if there is a remaining reference core content related to the reference core content, it is determined that the reference core content meets a preset condition; if there is no remaining reference core content related to the reference core content, it is determined that the reference core content does not meet the preset condition; the reference core content meeting the preset condition is used as the core content of the title of the to-be-recommended album.

2. The method of claim 1, wherein, The step of obtaining the reference core content of the title of the to-be-recommended album and the type of the reference core content based on the search text in each historical click record comprises: obtaining each pending core content of the title of the to-be-recommended album according to the title of the to-be-recommended album and the search text in each historical click record; processing the search text in each historical click record by using a pre-constructed dictionary and a pre-trained entity extraction model to obtain at least one entity set; based on each pending core content and each entity set, obtaining the reference core content and the type of the reference core content.

3. The method of claim 2, wherein, The step of obtaining at least one pending core content of the title of the to-be-recommended album according to the title of the to-be-recommended album and the search text in each historical click record comprises: de-duplicating the search text in the plurality of historical click records to obtain at least one candidate text; for each candidate text, the ratio of the number of historical click records in which the search text is the candidate text to the total number of historical click records is used as the album click rate corresponding to the candidate text, to obtain the album click rate corresponding to each candidate text; determining each target text from all candidate texts according to the album click rate; the overlapping part of each target text and the title of the to-be-recommended album is used as one pending core content.

4. The method of claim 2, wherein, Each entity set has a corresponding entity type. The step of obtaining the reference core content and the type of the reference core content based on each pending core content and each entity set comprises: for each pending core content, if there is a target set in all entity sets, the pending core content is used as one reference core content, and the entity type corresponding to the target set is used as the type of the reference core content, wherein the target set has an entity with the same name as the pending core content; if there is no target set in all entity sets, the pending core content is retained. If none of the core contents can be the reference core content, the title of the album to be recommended is taken as the reference core content, and a preset type is set as the type of the reference core content.

5. A search request processing method characterized by comprising: The method comprises: obtaining a user search request, the user search request comprising search text input by a user; calculating similarity between the search text and core content of a title of each album to be pushed, wherein the core content of the title of the album to be pushed is obtained by using the title core content determination method according to any one of claims 1-4; pushing each of the albums to be pushed in sequence according to the size of the similarity to respond to the user search request.

6. A title core content determination apparatus characterized by comprising: The device comprises: a first obtaining module configured to obtain a plurality of historical click records of an album to be recommended within a preset time range, wherein each of the historical click records comprises search text input by a user; a first processing module configured to: obtain reference core content of a title of the album to be recommended and a type of the reference core content based on the search text in each of the historical click records; if the reference core content is multiple and the type of each of the reference core contents is different, determine a relationship between each of the reference core contents and the remaining reference core contents by using a pre-constructed knowledge graph; for each of the reference core contents, if there is remaining reference core content related to the reference core content, determine that the reference core content meets a preset condition; if there is no remaining reference core content related to the reference core content, determine that the reference core content does not meet the preset condition; take the reference core content meeting the preset condition as core content of the title of the album to be recommended.

7. A search request processing apparatus characterized by comprising: The device comprises: a second obtaining module configured to obtain a user search request, the user search request comprising search text input by a user; a second processing module configured to calculate similarity between the search text and core content of a title of each album to be pushed, wherein the core content of the title of the album to be pushed is obtained by using the title core content determination method according to any one of claims 1-4; a pushing module configured to push each of the albums to be pushed in sequence according to the size of the similarity to respond to the user search request.

8. A computer device, comprising: It comprises a memory and a processor, the memory stores a computer program, and the processor implements the title core content determination method according to any one of claims 1-4 and / or the search request processing method according to claim 5 when executing the computer program.

9. A computer-readable storage medium, characterized in that, It stores a computer program, and the computer program is executed by a processor to implement the title core content determination method according to any one of claims 1-4 and / or the search request processing method according to claim 5.

Images