Information list recall method and device, computer storage medium, and electronic device

Abstract

The embodiments of the present disclosure relate to the field of computer technology, and more specifically, the embodiments of the present disclosure relate to an information list recall method and apparatus, a computer storage medium, and an electronic device. The method includes: responding to a search request, obtaining a search intent corresponding to the search request; when the search intent is a first type of search intent, segmenting the search text to obtain a first segmentation result, and obtaining a first target information list including the first segmentation result; when the search intent is a second type of search intent, performing entity linking based on the search text to obtain a target information entity corresponding to the search text, and obtaining a second information list including the target information entity; segmenting the search text to obtain a second segmentation result, and obtaining a third information list including the second segmentation result; and obtaining a second target information list based on the second information list and the third information list. This method improves the diversity of the recalled information list.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of computer technology, and more specifically, to a method and apparatus for recalling an information list, a computer storage medium, and an electronic device. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims, and no statement herein is admitted to be prior art by inclusion in this section.

[0003] Recalling a list of information is a fundamental feature of many apps. When a user enters a search term in the app, the app searches for the information, retrieves a list of information corresponding to the search term, and then recalls the list. For music apps, the list of information recalled is the playlist corresponding to the search term; for books, the list of information recalled is the book list corresponding to the search term; for movies, the list of information recalled is the movie list corresponding to the search term, and so on. Summary of the Invention

[0004] However, in the current information list recall method, on the one hand, the application can only recall based on the literal search text entered by the user, and the recall results are single and difficult to meet the needs of users; on the other hand, traditional information list recall is based on text matching of the information list name, that is, the user's search text needs to be included in the name of the information list. However, when performing text matching based on the information list name, the information list is usually a collection of information entities centered on a certain topic. When creating an information list, the entity list is usually not named after the name of the information entity, resulting in a single recall result; on the other hand, when searching, the user expects to recall an information list containing the search entity corresponding to the search text, rather than just an information list containing the search text in the name of the information list, resulting in the recall result not meeting the user's needs.

[0005] Therefore, there is a great need for an improved information list recall method and device, computer storage medium and electronic device, so as to provide an information list recall method that can improve the diversity of the recalled information list.

[0006] In this context, embodiments of the present disclosure are intended to provide an information list recall method and apparatus, a computer storage medium, and an electronic device.

[0007] According to one aspect of the present disclosure, a method for recalling an information list is provided, comprising:

[0008] In response to a search request, inputting a search text corresponding to the search request into a preset intent classification model to obtain a search intent corresponding to the search request;

[0009] When the search intent is a first type of search intent, segmenting the search text to obtain a first segmentation result, and obtaining a first target information list whose name includes the first segmentation result;

[0010] When the search intent is the second type of search intent, entity linking is performed based on the search text to obtain a target information entity corresponding to the search text, and a second information list including the target information entity is obtained in the information list;

[0011] Performing word segmentation on the search text to obtain a second word segmentation result, and obtaining a third information list whose name includes the second word segmentation result;

[0012] A second target information list is obtained according to the second information list and the third information list.

[0013] According to one aspect of the present disclosure, there is provided an information list recall device, comprising:

[0014] An intent analysis module is configured to respond to a search request, input a search text corresponding to the search request into a preset intent classification model, and obtain a search intent corresponding to the search request;

[0015] A first target information list acquisition module is configured to, when the search intent is a first type of search intent, segment the search text to obtain a first segmentation result, and acquire a first target information list whose name includes the first segmentation result;

[0016] an entity linking module, configured to, when the search intent is a second type of search intent, perform entity linking based on the search text, obtain a target information entity corresponding to the search text, and obtain a second information list including the target information entity in the information list;

[0017] an entity search module, configured to segment the search text to obtain a second segmentation result, and obtain a third information list whose name includes the second segmentation result;

[0018] The second target information list acquisition module is configured to obtain a second target information list according to the second information list and the third information list.

[0019] According to one aspect of the present disclosure, a computer storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned information list recall method is implemented.

[0020] According to one aspect of the present disclosure, there is provided an electronic device, including:

[0021] processor; and

[0022] a memory for storing executable instructions of the processor;

[0023] The processor is configured to execute any one of the above-mentioned information list recall methods by executing the executable instructions.

[0024] According to the information list recall method of the embodiment of the present disclosure, on the one hand, after obtaining the search request, the search text corresponding to the search request is input into the preset intent classification model, and the search request is classified by the preset intent classification model to obtain the search intent. After obtaining the search intent, the search request is processed accordingly according to the type of the search intent, thereby improving the efficiency of information list recall; on the other hand, when the search intent is the second type of search intent, not only can the second text be segmented to obtain the second segmentation result, but also an information list including the second segmentation result in the name of the information list obtained from the second segmentation result can be obtained, and entity linking can be performed according to the search text to obtain the target information entity corresponding to the search text, thereby improving the accuracy of information list recall; after obtaining the target information entity, an information list including the target information entity in the information list is obtained through the target information entity, thereby improving the richness of the recall results. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the detailed description below with reference to the accompanying drawings, in which several embodiments of the present disclosure are shown by way of example and not limitation, wherein:

[0026] Figure 1 The following schematically shows a flow chart of a method for recalling an information list according to an embodiment of the present disclosure;

[0027] Figure 2 The following schematically shows a block diagram of an information list recall system according to an embodiment of the present disclosure;

[0028] Figure 3 A flowchart schematically illustrates a method for inputting a search text corresponding to a search request into a preset intent classification model to obtain a search intent corresponding to the search request according to an embodiment of the present disclosure;

[0029] Figure 4 Schematically shows a schematic diagram of a preset intent classification model according to an embodiment of the present disclosure;

[0030] Figure 5 A flowchart of a method for obtaining a first target information list including a first word segmentation result in the name of the information list according to an embodiment of the present disclosure is schematically shown;

[0031] Figure 6 A flowchart schematically illustrates a method for performing entity linking based on a search text to obtain a target information entity corresponding to the search text when the search intent is the first search intent in the second type of search intent according to an embodiment of the present disclosure;

[0032] Figure 7 A flowchart of a method for screening a first information entity to obtain a target information entity according to an embodiment of the present disclosure is schematically shown;

[0033] Figure 8 A flowchart schematically illustrates a method for performing entity linking based on a search text to obtain a target information entity corresponding to the search text when the search intent is a second search intent in a second type of search intent according to an embodiment of the present disclosure;

[0034] Figure 9 A flowchart of a method for screening a second information entity to obtain a target information entity according to an embodiment of the present disclosure is schematically shown;

[0035] Figure 10 A flowchart of a method for obtaining a second information list including a target information entity in an information list according to an embodiment of the present disclosure is schematically shown;

[0036] Figure 11 The following schematically shows a schematic diagram of an initial information list according to an embodiment of the present disclosure;

[0037] Figure 12 A flowchart of a method for searching a target information entity in an entity inverted index to obtain a second information list including the target information entity in an initial information list according to an embodiment of the present disclosure is schematically shown;

[0038] Figure 13 A flowchart schematically illustrates a method for segmenting a search text, obtaining a second segmentation result, and acquiring a third information list whose name includes the second segmentation result according to an embodiment of the present disclosure;

[0039] Figure 14 Schematically shows a flow chart of a method for obtaining a second target information list based on a second information list and a third information list according to an embodiment of the present disclosure;

[0040] Figure 15The following schematically shows a flow chart of a method for recalling an information list according to an embodiment of the present disclosure;

[0041] Figure 16 A schematic diagram of a computer storage medium according to an embodiment of the present disclosure is shown;

[0042] Figure 17 A block diagram of an electronic device according to a disclosed embodiment is schematically shown.

[0043] In the drawings, the same or corresponding reference numerals denote the same or corresponding parts. DETAILED DESCRIPTION

[0044] The principles and spirit of the present disclosure will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided solely to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. Rather, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0045] Those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present disclosure may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software.

[0046] According to an embodiment of the present disclosure, an information list recall method, an information list recall device, a computer storage medium, and an electronic device are provided.

[0047] In this document, any number of elements in the drawings is for illustration and not for limitation, and any naming is for distinction only and does not have any limiting meaning.

[0048] The principles and spirit of the present disclosure are described in detail below with reference to several representative embodiments of the present disclosure. SUMMARY OF THE INVENTION

[0050] With the existing information list recall, on the one hand, the application responds to the user's search request, directly queries the information database according to the search text corresponding to the user's search request, obtains the information list in the information database whose information list name includes the search text, and recalls the information list whose information list name includes the search text; on the other hand, the information list that the user wants to recall is the information list containing the information entity corresponding to the search text, rather than just the information list whose name includes the search text; on the other hand, when the user creates an information list in the application, he seldom names the information list with the name of any information entity in the information list; the above three aspects all lead to the fact that the recall results in the related technology include fewer information lists, and the recalled information lists are difficult to meet the needs of users.

[0051] In view of the above content, the basic idea of the present disclosure is: according to the information list recall method and information list recall device implemented in the present disclosure, first, in response to the user's search request, the search text corresponding to the search request is input into a preset intent classification model to obtain the search intent; then, the type of the search intent is determined. When the search intent is a first type of search intent, the search text is segmented to obtain a first segmentation result, and a first target information list including the first segmentation result in the name of the information list is obtained; when the search intent is a second type of search intent, the search text can be segmented to obtain a second segmentation result, and a third target information list including the second segmentation result in the name of the information list is obtained. Entity linking can also be performed based on the search text to obtain a target information entity corresponding to the search text, a second information list including the target information entity in the information list is obtained, and a second Mu Niao information list is obtained based on the second information list and the third information list.

[0052] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0053] Exemplary Methods

[0054] First, combine Figure 1 The information list recall method according to an exemplary embodiment of the present disclosure will be described.

[0055] refer to Figure 1 , the information list recall method may include the following steps:

[0056] S110: In response to a search request, input a search text corresponding to the search request into a preset intent classification model to obtain a search intent corresponding to the search request;

[0057] S120: When the search intent is a first type of search intent, perform word segmentation on the search text to obtain a first word segmentation result, and obtain a first target information list whose name includes the first word segmentation result;

[0058] S130: When the search intent is a second type of search intent, perform entity linking based on the search text to obtain a target information entity corresponding to the search text, and obtain a second information list including the target information entity in the information list;

[0059] S140: Segment the search text to obtain a second segmentation result, and obtain a third information list whose name includes the second segmentation result;

[0060] S150. Obtain a second target information list according to the second information list and the third information list.

[0061] In the information list recall method of the embodiment of the present disclosure, on the one hand, after obtaining the search request, the search text corresponding to the search request is input into the preset intent classification model, and the search request is classified by the preset intent classification model to obtain the search intent. After obtaining the search intent, the search request is processed accordingly according to the type of search intent, thereby improving the efficiency of information list recall; on the other hand, when the search intent is the second type of search intent, not only can the second text be segmented to obtain the second segmentation result, and the information list including the second segmentation result in the name of the information list obtained from the second segmentation result can be obtained, but also the search text can be entity linked to obtain the target information entity corresponding to the search text, thereby improving the accuracy of information list recall through the entity link; after obtaining the target information entity, the information list including the target information entity in the information list is obtained through the target information entity, thereby improving the richness of the recall result. Among them, the information list can be a playlist list, a video list, an audio list, a text list, etc., and the information list is not specifically limited in this example embodiment.

[0062] Secondly, combined Figure 2 To describe the information list recall system involved in the exemplary embodiment of the present disclosure. Figure 2As shown, the information list recall system may include: an intent recognition model 210, a word segmentation module 220, an entity link module 230, an engine recall module 240 and a result fusion module 250. Among them, the intent recognition model 210 is used to classify the search text corresponding to the user's search request to obtain the user's search intent, wherein the search intent can be a first type of search intent or a second type of search intent; the word segmentation module 220 is used to segment the search text corresponding to the first type of search intent to obtain a first word segmentation result; and to segment the search text corresponding to the second type of search intent to obtain a second word segmentation result; the entity linking module 230 is used to perform entity linking based on the search text when the search intent is the second type of search intent to obtain a target information entity corresponding to the search text; the engine recall module 240 is used to obtain a first target information list whose name includes the first word segmentation result in the information list when the search intent is the first type of search intent; or to obtain a third information list whose name includes the second word segmentation result in the information list and a second information list including the target information entity in the information list when the search intent is the second type of search intent; the result fusion module 250 is used to merge and deduplicate the second information list and the third information list when the search intent is the second type of search intent to obtain a second target information list.

[0063] The following will be combined Figure 2 Steps S110 to S150 of the present disclosure are explained and illustrated in detail.

[0064] In step S110 , in response to a search request, a search text corresponding to the search request is input into a preset intent classification model to obtain a search intent corresponding to the search request.

[0065] In this example embodiment, in response to a user's search request, after receiving the user's search request, the user's search request is input into a preset intent classification model, and the search intent corresponding to the user's search request is obtained through the preset intent classification model. Among them, the preset intent classification model can be an intent classification model based on BERT (Bidirectional Encoder Representations from Transformers, a bidirectional encoder of transformers). In this BERT, each input embedding is composed of three embeddings, namely position embeddings, segment embeddings, and token embeddings.

[0066] In this example embodiment, reference Figure 3As shown, inputting the search text corresponding to the search request into a preset intent classification model to obtain the search intent corresponding to the search request may include:

[0067] Step S310: performing word segmentation on the search text to obtain a third word segmentation result, adding a first tag to the head of the third word segmentation result and a second tag to the tail of the third word segmentation result to obtain target input data;

[0068] Step S320: input the target input data into the preset intention classification model, and extract text features of the target input data through the preset intention classification model;

[0069] Step S330: Acquire a feature corresponding to the first label from the text features, and determine the feature corresponding to the first label as a semantic vector of the target input data;

[0070] Step S340: perform intent classification on the semantic vector of the target input data to obtain a search intent corresponding to the search request.

[0071] Below, steps S310 to S340 will be further explained and illustrated. Figure 4 The preset intent classification model shown in the figure may include training data, BERT, a fully connected layer, a softmax layer, and an output layer. First, the training data is constructed. The training data mainly includes manually labeled search intents corresponding to the user's search requests. However, the amount of manually labeled search intent data corresponding to the user's search requests is relatively small. Therefore, the information library can also be associated with the user's search request to construct a large amount of training data. For example, in a music application, when the user's search text is "JJ Lin", the corresponding singer can be directly matched in the music library, and the singer "JJ Lin" can be marked with the classification label of "artist".

[0072] After obtaining the training data, the BERT model can be trained using the training data to obtain the weight parameters of the trained BERT model, and the preset intent classification model can be obtained using the weight parameters. The loss function of the model can be a cross-entropy loss function or other loss function. In this example embodiment, the loss function is not specifically limited. After obtaining the preset intent classification model, the search text corresponding to the user's search request can be segmented into single words to obtain a third segmentation result, and the first label CLS is added to the head of the third segmentation result and the second label SEP is added to the tail to obtain the target input data; then, the target input data is used as input and input into the trained BERT model, and the text features of the target input data are extracted through the trained BERT model; then, the features corresponding to the first label in the text features of the target input data are obtained, that is, the text features of the CLS label part are obtained, and the text features of the CLS label part are determined as the semantic vector of the target input data; finally, the semantic vector of the target input data is classified using the fully connected layer, and the output vector is obtained through the softmax layer. The output vector corresponds to the probability of each search intent, and the search intent with the highest probability is determined as the search intent corresponding to the user's search request.

[0073] For example, in a music scenario, the search intent corresponding to a user's search request can be any of artist, tag, song, and lyrics; in a movie scenario, the search intent corresponding to a user's search request can be any of artist, tag, movie, and lines; in a book scenario, the search intent corresponding to a user's search request can be any of author, tag, book, and quote. In this example embodiment, the search intent for each scenario is not limited. When there are four search intents, a fully connected layer can be used to classify the semantic vector of the target input data into four categories, and a softmax layer can be used to obtain a four-dimensional output vector. In this example embodiment, the types of search intent are not specifically limited.

[0074] In step S120, when the search intent is a first type of search intent, the search text is segmented to obtain a first segmentation result, and a first target information list whose name includes the first segmentation result is obtained.

[0075] In this example embodiment, in the music scenario, if the user's search intent is for artists or tags; in the movie scenario, if the user's search intent is for artists or tags; in the book scenario, if the user's search intent is for writers or tags, the user's search intent can be determined as the first type of search intent. In the music scenario, if the user's search intent is for songs or lyrics; in the movie scenario, if the user's search intent is for movies or lines; in the book scenario, if the user's search intent is for books or famous quotes, the user's search intent can be determined as the second type of search intent.

[0076] In this example embodiment, in a music scenario, when the search intent corresponding to the user's search request is a first type of search intent, that is, when the user's search intent is an artist or a label, the search text corresponding to the user's search request can be segmented to obtain a first segmentation result, and then a first target information list in the information library is obtained based on the first segmentation result. Figure 5 As shown, obtaining a first target information list including the first word segmentation result in the name of the information list may include:

[0077] Step S510: Obtain a text inverted index corresponding to the information database containing the information list;

[0078] Step S520: Search the text inverted index for the segmented words included in the first segmented word result to obtain the first target information list, wherein the name of the first target information list includes all the segmented words in the first segmented word result.

[0079] Steps S510 and S520 are further explained and illustrated below. Specifically, after the search text corresponding to the user's search request is segmented to obtain a first segmentation result, a text inverted index of the information library can be obtained. The information library includes multiple information lists, and the text inverted index is constructed based on the multiple information lists. Then, all the segmented words included in the first segmentation result are searched in the text inverted index to obtain a first target information list, where the name of the first target information list includes all the segmented words in the first segmentation result. The text inverted index is a data structure that all search engines that support full-text search rely on. It is used to store the mapping of the storage location of a word in a document or a group of documents during a full-text search. Based on the index structure, a given word can be retrieved to obtain a list of documents containing the word. In different scenarios, the library represented by the information library is different. In the music scenario, the information library can be a music library; in the data scenario, the information library can be a book library; in the movie scenario, the information library can be a film library. In this example embodiment, the information library is not specifically limited.

[0080] For example, when the search text corresponding to the user's search request is "Jay Chou's good songs", the preset intent recognition model recognizes that the user's search intent is the artist, and the search text can be segmented, and the search text "Jay Chou's good songs" can be segmented into "Jay Chou / good-sounding / of / songs". After the segmentation results are obtained, the segmentation result Jay Chou / good-sounding / of / songs can be searched in the text inverted index of the information library, and the first target information list in the information library is obtained, in which the name of the information list includes the four segmentations Jay Chou, good-sounding, of, and songs. The first target information list is a playlist including the four segmentations, that is, in the obtained playlist, the name of the playlist can be exactly the same as the user's search text, or the playlist name can include the four segmentations. This is not specifically limited in this example embodiment. In this example embodiment, the retrieval efficiency is improved by the text inverted index of the text library.

[0081] In step S130, when the search intent is a second type of search intent, entity linking is performed based on the search text to obtain a target information entity corresponding to the search text, and a second information list including the target information entity is obtained in the information list.

[0082] In this example embodiment, in a music scenario, when the search intent corresponding to the user's search request is a second type of search intent, that is, when the user's search intent is a song or lyrics, entity linking can be performed based on the user's search text to obtain a target information entity corresponding to the user's search text, that is, a target song corresponding to the user's search request; after obtaining the target information entity, a second information list including the target information entity can be obtained from the information list in the information library based on the information entity. Among them, an entity is a specific actual object. When in a music scenario, an entity can be a unique identifier of any artist or any song; when in a book scenario, an entity can be a unique identifier of any writer or any book; when in a movie scenario, an entity can be any actor or director, or a unique identifier of any movie. In this example embodiment, there is no specific limitation on the entity. Entity linking refers to associating the user's search text with an entity.

[0083] refer to Figure 6 As shown, when the search intent is the first search intent in the second type of search intent, performing entity linking based on the search text to obtain a target information entity corresponding to the search text may include:

[0084] Step S610: Recognize the search text through named entity recognition to obtain entity text included in the search text;

[0085] Step S620: Match the entity text in an information database to obtain a first information entity corresponding to the entity text and included in the information database, wherein the name of the first information entity is the entity text;

[0086] Step S630: Screen the first information entity to obtain the target information entity.

[0087] Below, step S610-step S630 will be further explained and illustrated. Specifically, in a music scenario, the first search intent in the second type of search intent can be a song; in a movie scenario, the first search intent in the second type of search intent can be a movie; in a book scenario, the first search intent in the second type of search intent can be a book, and this is not specifically limited in this example embodiment. When the search intent is the first search intent in the second type of search intent, the search text can be identified by named entity recognition to obtain the entity text included in the search text; then the entity text is used to match in the information database to obtain the first information entity whose entity name is the entity text included in the information database. Among them, named entity recognition (NER) refers to the identification of entities with specific meanings in the text, mainly including names of people, places, organizations, proper nouns, etc., as well as text such as time, quantity, currency, proportional values, etc. In the music scenario, named entity recognition can obtain entity names such as songs or artists.

[0088] For example, in a music scenario, when the user's search text is "Jay Chou's Qilixiang", the preset intent recognition model determines that the user's search intention is a song, and through named entity recognition, the entity text in the search text is obtained as "Qilixiang"; then, the entity text "Qilixiang" is matched in the information database to obtain the song named Qilixiang in the information database.

[0089] In this exemplary embodiment, after obtaining a first information entity named entity text in the information database, since the information database may include multiple information entities named entity text, the obtained first information entity can be screened. Figure 7 The filtering of the first information entity to obtain the target information entity may include:

[0090] Step S710: Obtain a first matching degree, a first popularity score, and a first consumption score of the first information entity with the search text;

[0091] Step S720: Obtain a first comprehensive score for the first information entity based on the first matching degree, the first popularity score, and the first consumption score;

[0092] Step S730: Filter the first information entities according to the first comprehensive score to obtain the target information entity.

[0093] Below, steps S710 to S730 will be further explained and illustrated. Specifically, after obtaining multiple first information entities, the first matching degree, first popularity score, and first consumption score of each first information entity with the search text can be obtained; after obtaining the first matching degree, first popularity score, and first consumption score of the first information entity, the first matching score, first popularity score, and first consumption score can be calculated according to the preset weights of the above three dimensions to obtain the first comprehensive score of the first information entity; finally, the first information entities can be screened by the first comprehensive score, and the top two information entities with the highest first comprehensive scores can be determined as the target information entities.

[0094] Among them, the first matching degree is the degree of match between the name of the first information entity and the search text. Different scores can be assigned to the first matching degree according to the degree of match between the name of the first information entity and the search text. In this example embodiment, the first matching degree score is not specifically limited. The first popularity score is a measure of the quality of the first information entity. It can be based on the effective playback volume of the first information entity within a preset time period or the level of the artist corresponding to the first information entity. The first popularity score of the first information entity can also be measured based on factors such as user comments, reposts, and likes on the first information entity. In this example embodiment, the measurement factors of the first popularity score are not specifically limited. The first consumption score is the effective playback rate of the first information entity under the search request. The effective playback means that the playback time of the first information entity exceeds the preset time. In this example embodiment, the preset time is not specifically limited. The first consumption score of the first information entity can be obtained by combining the effective playback volume of the first information entity under the search request and the exposure volume of the first information entity under the search request, that is, the first consumption score = the effective playback volume of the first information entity under the search request / the exposure volume of the first information entity under the search request.

[0095] In this exemplary embodiment, in reference to Figure 8 As shown, when the search intent is the second search intent in the second type of search intent, performing entity linking based on the search text to obtain a target information entity corresponding to the search text may include:

[0096] Step S810: Acquire a first type of information entity and entity content corresponding to the first type of information entity in the information database, and establish an entity content index based on the first type of information entity and the entity content;

[0097] Step S820: Search the search text in the entity content index to obtain a second information entity corresponding to the search text;

[0098] Step S830: Screen the second information entity to obtain the target information entity.

[0099] Below, step S810-step S830 will be further explained and illustrated. In a music scenario, the second search intent in the second type of search intent can be lyrics; in a movie scenario, the second search intent in the second type of search intent can be lines; in a book scenario, the first search intent in the second type of search intent can be a famous quote, which is not specifically limited in this example embodiment. When the search intent is the second search intent in the second type of search intent, the first type of information entity in the information library and the entity content corresponding to the first type of information entity can be obtained, and an entity content index can be established based on the first type of information entity and the entity content corresponding to it; after the entity content index of the first type of information entity in the information library is constructed, the search text can be searched based on the entity content index to obtain the second information entity including the search text in the entity content; finally, the obtained second information entity is screened to obtain the target information entity. In a music scenario, the first type of information entity may be a popular song, which is a song with a high number of hits within a preset time period, and the entity content corresponding to the first type of information entity may be the lyrics of the song; in a movie scenario, the first type of information entity may be a popular movie, which is a movie or TV series with a high number of views within a preset time period, and the entity content corresponding to the first type of information entity may be lines from the movie; in a book scenario, the first type of information entity may be a popular book, which is a book with a high number of views within a preset time period, and the entity content corresponding to the first type of information entity may be sentences from the book. The preset time period for popularity may be within 24 hours, within 7 days, or within 30 days, and the preset time period is not specifically limited in this example embodiment.

[0100] For example, when in a music scene, the search text entered by the user is "Who says only those who stand in the light are heroes?", the preset intent recognition model is used to obtain the user's search intention as lyrics, first obtain the popular songs in the music library, and then build an entity content index based on the popular songs and the lyrics included in the popular songs. After the entity content index is built, the entity content index is used to search for "Who says only those who stand in the light are heroes", and a second information entity is obtained, including the search text in the entity content, that is, the lyrics "Who says only those who stand in the light are heroes".

[0101] After obtaining the second information entity, refer to Figure 9As shown, screening the second information entity to obtain the target information entity may include:

[0102] Step S910: Obtain a second matching degree, a second popularity score, and a second consumption score of the second information entity with the search text;

[0103] Step S920: Obtain a second comprehensive score for the second information entity based on the second matching degree, the second popularity score, and the second consumption score;

[0104] Step S930: Filter the second information entities according to the second comprehensive score to obtain the target information entity.

[0105] Below, steps S910 to S930 will be further explained and illustrated. Specifically, after obtaining multiple second information entities, the second matching degree, second popularity score, and second consumption score of each second information entity with the search text can be obtained; after obtaining the second matching degree, second popularity score, and second consumption score of the second information entity, the second matching score, second popularity score, and second consumption score can be calculated according to the preset weights of the above three dimensions to obtain the second comprehensive score of the second information entity; finally, the second information entities can be screened by the second comprehensive score, and the top two information entities with the highest second comprehensive scores can be determined as the target information entities.

[0106] Among them, the second matching degree is the degree of matching between the second information entity and the search text. Different scores can be assigned to the second matching degree according to the degree of matching between the name of the second information entity and the search text. In this example embodiment, the second matching degree score is not specifically limited. The second popularity score is a measure of the quality of the second information entity. It can be based on the effective playback volume of the second information entity within a preset time period or the level of the artist corresponding to the second information entity. The second popularity score of the second information entity can also be measured based on factors such as user comments, reposts, and likes on the second information entity. In this example embodiment, the measurement factors of the second popularity score are not specifically limited. The second consumption score is the effective playback rate of the second information entity under the search request. The effective playback means that the playback time of the second information entity exceeds the preset time. In this example embodiment, the preset time is not specifically limited. The second consumption score of the second information entity can be obtained by combining the effective playback volume of the second information entity under the search request and the exposure volume of the second information entity under the search request, that is, the second consumption score = the effective playback volume of the second information entity under the search request / the exposure volume of the second information entity under the search request.

[0107] In this exemplary embodiment, after obtaining the target information entity corresponding to the search text, a second information list including the target information entity can be obtained. Figure 10 As shown, obtaining the second information list including the target information entity in the information list may include:

[0108] Step S1010: Obtain all information lists in the information database, and filter all the information lists to obtain an initial information list;

[0109] Step S1020: Acquire information entities included in the initial information list, and construct an entity inverted index based on the information entities included in the initial information list;

[0110] Step S1030: Search the target information entity in the entity inverted index to obtain a second information list including the target information entity in the initial information list.

[0111] Below, step S1010 to step S1030 will be further explained and illustrated. Specifically, after obtaining the target information entity, all information lists included in the information library can be obtained. In the music scene, all information lists are all playlists; in the movie scene, all information lists are all movie lists; in the book scene, all information lists are all book lists; after obtaining all information lists, all obtained information lists are filtered to obtain an initial information list. After obtaining the initial information list, the information entities and information of the information entities included in the initial information list can be obtained, and an entity inverted index can be constructed based on the information of the initial information list and the information of the information entities included in the initial information list; finally, the target information entity is searched through the entity inverted index to obtain a second information list including the target information entity in the initial information list.

[0112] Among them, the information list can be filtered according to the number of times the information list is played, the number of times the information list is collected, and the attributes of the information entities in the information list. When the number of times the information list is played, the number of times the information list is collected, and the attribute information of the information entity all meet the conditions, the information list can be determined as the initial information list; for example, when the number of times the information list is played reaches more than 10,000, the number of times the information list is collected reaches more than 10, and the information entities included in the information list cannot all be non-copyrighted information entities, in this example embodiment, there is no specific limitation on the conditions that the information list needs to meet; when constructing an entity inverted index, the information of the initial information list can be a unique identifier of the initial information list, a uniform resource locator of the initial information list, etc., and in this example embodiment, there is no specific limitation on the information of the information list; the information of the entity information can be a unique identifier of the information entity included in the information list.

[0113] refer to Figure 11 As shown, Figure 11 For any information list in the initial information list, after obtaining the information list, that is, after obtaining any initial playlist in the music library, an entity inverted index can be constructed based on the unique identifier of the playlist, the uniform resource locator of the playlist, and the unique identifiers of the songs included in the playlist. The generated entity inverted index can be:

[0114] Playlist unique identifier: 123456; Uniform Resource Locator: http: / / XXXXXX

[0115] Song list: ["1952728254","209073","209076","209079"]

[0116] In this example embodiment, reference Figure 12 As shown, searching the target information entity in the entity inverted index to obtain a second information list including the target information entity in the initial information list may include:

[0117] Step S1210: Search the target information entity in the entity inverted index to obtain an information list including the target information entity in the initial information list;

[0118] Step S1220: Obtain the correlation between the target information entity and the information list including the target information entity in the initial information list, and filter the information list including the target information entity in the initial information list according to the correlation to obtain the second information list.

[0119] Steps S1210 and S1220 will be further explained and illustrated below. Specifically, after obtaining the entity inverted index of the information entity included in the initial information list, the target information entity can be searched in the entity inverted index to obtain an information list in the initial list that includes the unique identifier of the target information entity. There may be multiple information lists obtained that include the unique identifier of the target information entity. Therefore, the relevance between the target information entity and the information list in each information list can be obtained, and the information lists that include the target information entity can be filtered based on the relevance to obtain a second information list.

[0120] Among them, the correlation between the target information entity and the information list can be determined by the average playback rate of the target information entity in the information list, and the playback rate of the target information entity on each day can be obtained by the number of times the target information list is played and the number of effective playbacks each day. At the same time, the playback rate of the target information entity every day within a preset time period is counted to obtain the average playback rate of the target information entity; the correlation between the target information entity and the information list can also be calculated by semantic similarity. In this example embodiment, there is no specific limitation on the method for calculating the correlation.

[0121] In this example embodiment, entity linking is performed through the user's search text to obtain a target information entity corresponding to the user's search text. Then, the target information entity is retrieved in the information database to obtain an information list including the target information entity, which not only improves the accuracy of information list recall, but also improves the richness of the recalled information list.

[0122] In step S140, the search text is segmented to obtain a second segmentation result, and a third information list including the second segmentation result in the name of the information list is obtained.

[0123] In this example embodiment, when the user's search intent is the second type of search intent, not only can the target information entity corresponding to the user's search text be accurately determined and the target information entity be retrieved, but the user's search text can also be segmented and retrieved based on the literal meaning of the search text. Figure 13 As shown, performing word segmentation on the search text to obtain a second word segmentation result, and obtaining a third information list whose name includes the second word segmentation result may include:

[0124] Step S1310: Obtain a text inverted index corresponding to the information database containing the information list;

[0125] Step S1320: Search the text inverted index for the words included in the second word segmentation result to obtain the third information list, wherein the name of the third information list includes all the words in the second word segmentation result.

[0126] Steps S1310 and S1320 are further explained and illustrated below. Specifically, after segmenting the search text corresponding to the user's search request to obtain a second segmentation result, an inverted text index of an information library can be obtained, which includes multiple information lists. Then, all the segmentations included in the second segmentation result are searched in the inverted text index to obtain a third target information list, where the name of the third target information list includes all the segmentations in the second segmentation result.

[0127] In step S150, a second target information list is obtained according to the second information list and the third information list.

[0128] In this example embodiment, the second information list and the third information list are both retrieved when the user's search intent is the second type of search intent. Therefore, the second information list and the third information list can be merged. Figure 14 As shown, obtaining a second target information list according to the second information list and the third information list may include:

[0129] Step S1410: screening the second information list and adding annotations to the remaining second information list according to the target information entity;

[0130] Step S1420: Merge the remaining second information list and the third information list, and remove duplicates from the merged information list;

[0131] Step S1430: Sort the deduplicated information list according to a preset priority to obtain the second target information list.

[0132] Below, steps S1410 to S1430 will be further explained and illustrated. Specifically, after obtaining the second information list, the second information list can be screened again to obtain the remaining second information list, wherein screening the second information list again can include: screening the second information list again according to the consumption score of any information entity included in the second information list, that is, screening the second information list according to the effective playback rate of any information entity included in the second information list under the search request. In this example embodiment, screening the second information list by the consumption score of any information entity in the second information list can prevent the second information list from being of insufficient quality or low relevance, thereby improving the user experience.

[0133] After filtering the second information list and obtaining the remaining second information list, since the name of the remaining second information list may be completely different from the user's search text, it is necessary to add an annotation to the remaining second information list, that is, the target entity included in the list can be added in the display interface of the second information list, further improving the user's search experience. In this example embodiment, the remaining second information list and the remaining third information list are both information lists recalled through the second type of search intent. Since the remaining second information list and the third information list are recall results from different paths, there may be a situation where the same information list exists in the remaining second information list and the third information list. Therefore, when merging the remaining second information list and the remaining third information list, the information lists that exist in the remaining second information list and the third information list can be deduplicated first. After deduplication, the deduplicated information lists can be sorted according to a preset priority to obtain the second target information list. Among them, the preset priority can be that the priority of the information list recalled through the text inverted index is higher than the priority of the information list recalled through the entity inverted index. In this example, the preset priority is not specifically limited.

[0134] In summary, the method provided by the present disclosure can be applied to recalling information lists. After receiving a user's search request, the search text corresponding to the search request is input into a preset intent classification model to obtain the user's search intent. When the search intent is a first type of search intent, the user's search text is segmented to obtain a first segmentation result. All segmentations included in the first segmentation result are searched in the text inverted index of the information library to obtain a first target information list in the information library whose names include all segmentations in the first segmentation result.

[0135] When the search intent is the second type of search intent, entity linking is performed based on the user's search text to obtain the target information entity corresponding to the user's search text. Then, an entity inverted index of the information library is constructed, and the target information entity is retrieved in the entity inverted index to obtain a second information list in the information library that includes the target information entity. In addition, the user's search text is segmented to obtain a second segmentation result, and all the segmentations included in the second segmentation result are searched in the text inverted index of the information library to obtain a third information list in which the name of the information list in the information library includes all the segmentations in the second segmentation result. After obtaining the second information list and the third information list, the second information list and the third information list are deduplicated and merged to obtain the second target information list. This improves the richness of the recalled information list and also improves the user's search experience.

[0136] Exemplary devices

[0137] After introducing the information list recall method according to the exemplary embodiment of the present disclosure, Figure 15 An information list recall device according to an exemplary embodiment of the present disclosure will be described.

[0138] refer to Figure 15 As shown, the information list recall device 15 of the exemplary embodiment of the present disclosure may include: an intention analysis module 1510, a first target information list acquisition module 1520, an entity link module 1530, an entity search module 1540 and a second target information list acquisition module 1550; wherein:

[0139] The intent analysis module 1510 is configured to respond to a search request, input a search text corresponding to the search request into a preset intent classification model, and obtain a search intent corresponding to the search request;

[0140] A first target information list acquisition module 1520 is configured to, when the search intent is a first type of search intent, segment the search text to obtain a first segmentation result, and acquire a first target information list whose name includes the first segmentation result;

[0141] An entity linking module 1530 is configured to, when the search intent is a second type of search intent, perform entity linking based on the search text to obtain a target information entity corresponding to the search text, and obtain a second information list including the target information entity in the information list;

[0142] The entity search module 1540 is configured to segment the search text to obtain a second segmentation result, and obtain a third information list whose name includes the second segmentation result.

[0143] The second target information list acquisition module 1550 is configured to obtain a second target information list according to the second information list and the third information list.

[0144] According to an exemplary embodiment of the present disclosure, the intention analysis module includes:

[0145] a text segmentation module, configured to perform word segmentation on the search text to obtain a third word segmentation result, add a first label to the head of the third word segmentation result and add a second label to the tail of the third word segmentation result to obtain target input data;

[0146] A text feature acquisition module, configured to input the target input data into the preset intent classification model and extract text features of the target input data through the preset intent classification model;

[0147] a semantic vector determination module, configured to obtain a feature corresponding to the first label from the text features, and determine the feature corresponding to the first label as a semantic vector of the target input data;

[0148] The search intent determination module is used to perform intent classification on the semantic vector of the target input data to obtain the search intent corresponding to the search request.

[0149] According to an exemplary embodiment of the present disclosure, the first target information list acquisition module includes:

[0150] A first index acquisition module is used to obtain a text inverted index corresponding to the information library containing the information list;

[0151] The information list acquisition module searches the text inverted index for the segmented words included in the first segmented result to obtain the first target information list, wherein the name of the first target information list includes all the segmented words in the first segmented result.

[0152] According to an exemplary embodiment of the present disclosure, the entity link module includes:

[0153] A named entity recognition module, configured to recognize the search text through named entity recognition to obtain entity text included in the search text;

[0154] A first information entity acquisition module is configured to perform a match in an information database based on the entity text to obtain a first information entity corresponding to the entity text and included in the information database, wherein the name of the first information entity is the entity text;

[0155] A target information entity acquisition module is used to filter the first information entity to obtain the target information entity

[0156] According to an exemplary embodiment of the present disclosure, the target information entity acquisition module includes:

[0157] A first factor acquisition module, configured to acquire a first matching degree, a first popularity score, and a first consumption score of the first information entity with the search text;

[0158] a comprehensive score determination module, configured to obtain a first comprehensive score of the first information entity based on the first matching degree, the first popularity score, and the first consumption score;

[0159] An entity screening module is used to screen the first information entity according to the first comprehensive score to obtain the target information entity.

[0160] According to an exemplary embodiment of the present disclosure, the entity link module includes:

[0161] An entity content index establishment module, configured to obtain first-category information entities and entity content corresponding to the first-category information entities included in the information database, and establish an entity content index based on the first-category information entities and the entity content;

[0162] A second information entity determination module, configured to search the entity content index for the search text to obtain a second information entity corresponding to the search text;

[0163] The second information entity screening module is used to screen the second information entity to obtain the target information entity.

[0164] According to an exemplary embodiment of the present disclosure, the second information entity screening module further includes:

[0165] A second factor acquisition module, configured to acquire a second matching degree, a second popularity score, and a second consumption score of the second information entity with the search text;

[0166] a second comprehensive score determination module, configured to obtain a second comprehensive score of the second information entity according to the second matching degree, the second popularity score, and the second consumption score;

[0167] The target information entity determination module is configured to screen the second information entities according to the second comprehensive score to obtain the target information entity.

[0168] According to an exemplary embodiment of the present disclosure, the entity link module further includes:

[0169] An initial information list acquisition module is used to acquire all information lists in the information database, filter all the information lists, and obtain an initial information list;

[0170] An entity inverted index building module, configured to obtain information entities included in the initial information list and build an entity inverted index based on the information entities included in the initial information list;

[0171] The target information entity search module is configured to search for the target information entity in the entity inverted index to obtain a second information list including the target information entity in the initial information list.

[0172] According to an exemplary embodiment of the present disclosure, the target information entity search module includes:

[0173] An information list acquisition module, configured to search the target information entity in the entity inverted index and obtain an information list including the target information entity in the initial information list;

[0174] The second information list determining module is configured to obtain the correlation between the target information entity and the information list including the target information entity in the initial information list, and filter the information list including the target information entity in the initial information list according to the correlation to obtain the second information list.

[0175] According to an exemplary embodiment of the present disclosure, the entity search module further includes:

[0176] A second text inverted index acquisition module is used to acquire a text inverted index corresponding to the information library containing the information list;

[0177] A third information list acquisition module is configured to search the text inverted index for the words included in the second word segmentation result to obtain the third information list, wherein the name of the third information list includes all the words in the second word segmentation result.

[0178] According to an exemplary embodiment of the present disclosure, the second target information list acquisition module includes:

[0179] a second information list screening module, configured to screen the second information list and add annotations to the remaining second information list according to the target information entity;

[0180] an information list deduplication module, configured to merge the remaining second information list and the third information list, and dedupe the merged information list;

[0181] The information list sorting module is used to sort the deduplicated information list according to a preset priority to obtain the second target information list.

[0182] Since the functional modules of the information list recall device in the disclosed embodiment are the same as those in the disclosed embodiment of the information list recall method, they will not be described in detail here.

[0183] Exemplary Storage Media

[0184] After introducing the information list recall method and device according to the exemplary embodiment of the present disclosure, Figure 16 Computer storage media according to exemplary embodiments of the present disclosure are described.

[0185] refer to Figure 16As shown, a program product 1600 for implementing the above method according to an embodiment of the present disclosure is described. The program product 1600 may be a portable compact disk read-only memory (CD-ROM) and include program code, and may be run on a device such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0186] The program product may be implemented in any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0187] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0188] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0189] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, and the like, as well as conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0190] Exemplary electronic devices

[0191] After introducing the storage medium of the exemplary embodiment of the present disclosure, next, reference is made to Figure 17 An electronic device according to an exemplary embodiment of the present disclosure will be described.

[0192] Figure 17 The electronic device 1700 shown is merely an example and should not limit the functionality and scope of use of the embodiments of the present disclosure.

[0193] like Figure 17 As shown, electronic device 1700 is implemented as a general-purpose computing device. Components of electronic device 1700 may include, but are not limited to, the aforementioned at least one processing unit 1710, the aforementioned at least one storage unit 1720, a bus 1730 connecting various system components (including storage unit 1720 and processing unit 1710), and a display unit 1740.

[0194] The storage unit stores program codes, which can be executed by the processing unit 1710, so that the processing unit 1710 performs the steps described in the "Exemplary Method" section of the present disclosure according to various exemplary embodiments. For example, the processing unit 1710 can perform the following steps: Figure 1 Steps S110 to S150 shown in FIG.

[0195] The storage unit 1720 may include a volatile storage unit, such as a random access memory unit (RAM) 17201 and / or a cache memory unit 17202 , and may further include a read-only memory unit (ROM) 17203 .

[0196] The storage unit 1720 may also include a program / utility 17204 having a set (at least one) of program modules 17205, such program modules 17205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0197] The bus 1730 may include a data bus, an address bus, and a control bus.

[0198] Electronic device 1700 can also communicate with one or more external devices 1800 (e.g., a keyboard, pointing device, Bluetooth device, etc.), and such communication can be performed via input / output (I / O) interface 1750. Electronic device 1700 also includes a display unit 1740, which is connected to input / output (I / O) interface 1750 for display. Furthermore, electronic device 1700 can also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 1760. As shown, network adapter 1760 communicates with other modules of electronic device 1700 via bus 1730. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with electronic device 1700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0199] It should be noted that while the detailed description above mentions several modules or submodules of the audio playback device and audio sharing device, this division is merely exemplary and not mandatory. In practice, depending on the embodiments of the present disclosure, the features and functions of two or more units / modules described above may be embodied in a single unit / module. Conversely, the features and functions of a single unit / module described above may be further divided and embodied by multiple units / modules.

[0200] Furthermore, although the operations of the disclosed method are described in a particular order in the accompanying drawings, this does not require or imply that the operations must be performed in this particular order, or that all illustrated operations must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0201] Although the spirit and principles of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the specific embodiments disclosed, and the division into various aspects does not mean that the features in these aspects cannot be combined to benefit. Such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the appended claims.

Claims

1. A method for recalling an information list, characterized in that: include: In response to a search request, inputting a search text corresponding to the search request into a preset intent classification model to obtain a search intent corresponding to the search request; When the search intent is a first type of search intent, segmenting the search text to obtain a first segmentation result, and obtaining a first target information list whose name includes the first segmentation result; When the search intent is the second type of search intent, entity linking is performed based on the search text to obtain a target information entity corresponding to the search text, and a second information list including the target information entity is obtained in the information list; Performing word segmentation on the search text to obtain a second word segmentation result, and obtaining a third information list whose name includes the second word segmentation result; A second target information list is obtained according to the second information list and the third information list.

2. The information list recall method according to claim 1, characterized in that: Inputting the search text corresponding to the search request into a preset intent classification model to obtain the search intent corresponding to the search request includes: Performing word segmentation on the search text to obtain a third word segmentation result, adding a first label to the head of the third word segmentation result and a second label to the tail of the third word segmentation result to obtain target input data; Inputting the target input data into the preset intent classification model, and extracting text features of the target input data through the preset intent classification model; Acquire a feature corresponding to the first label from the text features, and determine the feature corresponding to the first label as a semantic vector of the target input data; The semantic vector of the target input data is subjected to intent classification to obtain a search intent corresponding to the search request.

3. The information list recall method according to claim 1, characterized in that: Acquiring a first target information list whose name includes the first word segmentation result includes: Obtaining a text inverted index corresponding to the information library containing the information list; In the text inverted index, the segmented words included in the first segmented word result are searched to obtain the first target information list, wherein the name of the first target information list includes all the segmented words in the first segmented word result.

4. The information list recall method according to claim 1, characterized in that: When the search intent is the first search intent in the second type of search intent, performing entity linking according to the search text to obtain a target information entity corresponding to the search text includes: Identify the search text through named entity recognition to obtain entity text included in the search text; Matching the entity text in an information database to obtain a first information entity corresponding to the entity text and included in the information database, wherein the name of the first information entity is the entity text; The first information entity is screened to obtain the target information entity.

5. The information list recall method according to claim 4, characterized in that: Screening the first information entity to obtain the target information entity includes: Obtaining a first matching degree, a first popularity score, and a first consumption score of the first information entity with the search text; Obtaining a first comprehensive score for the first information entity based on the first matching degree, the first popularity score, and the first consumption score; The first information entity is screened according to the first comprehensive score to obtain the target information entity.

6. The information list recall method according to claim 1, characterized in that: When the search intent is the second search intent in the second type of search intent, performing entity linking according to the search text to obtain a target information entity corresponding to the search text includes: Acquire a first type of information entity and entity content corresponding to the first type of information entity included in the information database, and establish an entity content index based on the first type of information entity and the entity content; Searching the search text in the entity content index to obtain a second information entity corresponding to the search text; The second information entity is screened to obtain the target information entity.

7. The information list recall method according to claim 6, characterized in that: Screening the second information entity to obtain the target information entity includes: Obtaining a second matching degree, a second popularity score, and a second consumption score of the second information entity with the search text; Obtaining a second comprehensive score for the second information entity based on the second matching degree, the second popularity score, and the second consumption score; The second information entity is screened according to the second comprehensive score to obtain the target information entity.

8. The information list recall method according to claim 4 or 6, characterized in that: Acquiring a second information list including the target information entity in the information list includes: Obtaining all information lists in the information database, screening all the information lists, and obtaining an initial information list; Acquire information entities included in the initial information list, and construct an entity inverted index according to the information entities included in the initial information list; The target information entity is searched in the entity inverted index to obtain a second information list including the target information entity in the initial information list.

9. The information list recall method according to claim 8, characterized in that: Searching the target information entity in the entity inverted index to obtain a second information list including the target information entity in the initial information list includes: Searching the target information entity in the entity inverted index to obtain an information list including the target information entity in the initial information list; The relevance between the target information entity and the information list including the target information entity in the initial information list is obtained, and the information list including the target information entity in the initial information list is filtered according to the relevance to obtain the second information list.

10. The information list recall method according to claim 1, characterized in that: Segmenting the search text to obtain a second segmentation result, and obtaining a third information list whose name includes the second segmentation result, including: Obtaining a text inverted index corresponding to the information library containing the information list; In the text inverted index, the segmented words included in the second segmented word result are searched to obtain the third information list, wherein the name of the third information list includes all the segmented words in the second segmented word result.

11. The information list recall method according to claim 1, characterized in that: According to the second information list and the third information list, a second target information list is obtained, including: screening the second information list and adding annotations to the remaining second information list according to the target information entity; Merging the remaining second information list and the third information list, and deduplicating the merged information list; The deduplicated information list is sorted according to a preset priority to obtain the second target information list.

12. An information list recall device, characterized in that: include: An intent analysis module is configured to respond to a search request, input a search text corresponding to the search request into a preset intent classification model, and obtain a search intent corresponding to the search request; A first target information list acquisition module is configured to, when the search intent is a first type of search intent, segment the search text to obtain a first segmentation result, and acquire a first target information list whose name includes the first segmentation result; an entity linking module, configured to, when the search intent is a second type of search intent, perform entity linking based on the search text, obtain a target information entity corresponding to the search text, and obtain a second information list including the target information entity in the information list; an entity search module, configured to segment the search text to obtain a second segmentation result, and obtain a third information list whose name includes the second segmentation result; The second target information list acquisition module is configured to obtain a second target information list according to the second information list and the third information list.

13. The information list recall device according to claim 12, characterized in that: The intention analysis module includes: a text segmentation module, configured to perform word segmentation on the search text to obtain a third word segmentation result, add a first label to the head of the third word segmentation result and add a second label to the tail of the third word segmentation result to obtain target input data; A text feature acquisition module, configured to input the target input data into the preset intent classification model and extract text features of the target input data through the preset intent classification model; a semantic vector determination module, configured to obtain a feature corresponding to the first label from the text features, and determine the feature corresponding to the first label as a semantic vector of the target input data; The search intent determination module is used to perform intent classification on the semantic vector of the target input data to obtain the search intent corresponding to the search request.

14. The information list recall device according to claim 12, characterized in that: The first target information list acquisition module includes: A first index acquisition module is used to obtain a text inverted index corresponding to the information library containing the information list; The information list acquisition module searches the text inverted index for the segmented words included in the first segmented result to obtain the first target information list, wherein the name of the first target information list includes all the segmented words in the first segmented result.

15. The information list recall device according to claim 12, characterized in that: The entity link module includes: A named entity recognition module, configured to recognize the search text through named entity recognition to obtain entity text included in the search text; A first information entity acquisition module is configured to perform a match in an information database based on the entity text to obtain a first information entity corresponding to the entity text and included in the information database, wherein the name of the first information entity is the entity text; The target information entity acquisition module is used to screen the first information entity to obtain the target information entity.

16. The information list recall device according to claim 15, characterized in that: The target information entity acquisition module includes: A first factor acquisition module, configured to acquire a first matching degree, a first popularity score, and a first consumption score of the first information entity with the search text; a comprehensive score determination module, configured to obtain a first comprehensive score of the first information entity based on the first matching degree, the first popularity score, and the first consumption score; An entity screening module is used to screen the first information entity according to the first comprehensive score to obtain the target information entity.

17. The information list recall device according to claim 12, characterized in that: The entity link module includes: An entity content index establishment module, configured to obtain first-category information entities and entity content corresponding to the first-category information entities included in the information database, and establish an entity content index based on the first-category information entities and the entity content; A second information entity determination module, configured to search the entity content index for the search text to obtain a second information entity corresponding to the search text; The second information entity screening module is used to screen the second information entity to obtain the target information entity.

18. The information list recall device according to claim 17, characterized in that: The second information entity screening module further includes: A second factor acquisition module, configured to acquire a second matching degree, a second popularity score, and a second consumption score of the second information entity with the search text; a second comprehensive score determination module, configured to obtain a second comprehensive score of the second information entity according to the second matching degree, the second popularity score, and the second consumption score; The target information entity determination module is configured to screen the second information entities according to the second comprehensive score to obtain the target information entity.

19. The information list recall device according to claim 15 or 17, characterized in that: The entity link module also includes: An initial information list acquisition module is used to acquire all information lists in the information database, filter all the information lists, and obtain an initial information list; An entity inverted index building module, configured to obtain information entities included in the initial information list and build an entity inverted index based on the information entities included in the initial information list; The target information entity search module is configured to search for the target information entity in the entity inverted index to obtain a second information list including the target information entity in the initial information list.

20. The information list recall device according to claim 19, characterized in that: The target information entity search module includes: An information list acquisition module, configured to search the target information entity in the entity inverted index and obtain an information list including the target information entity in the initial information list; The second information list determining module is configured to obtain the correlation between the target information entity and the information list including the target information entity in the initial information list, and filter the information list including the target information entity in the initial information list according to the correlation to obtain the second information list.

21. The information list recall device according to claim 12, characterized in that: The entity search module further includes: A second text inverted index acquisition module is used to acquire a text inverted index corresponding to the information library containing the information list; A third information list acquisition module is configured to search the text inverted index for the words included in the second word segmentation result to obtain the third information list, wherein the name of the third information list includes all the words in the second word segmentation result.

22. The information list recall device according to claim 12, characterized in that: The second target information list acquisition module includes: a second information list screening module, configured to screen the second information list and add annotations to the remaining second information list according to the target information entity; an information list deduplication module, configured to merge the remaining second information list and the third information list, and dedupe the merged information list; The information list sorting module is used to sort the deduplicated information list according to a preset priority to obtain the second target information list.

23. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the information list recall method according to any one of claims 1 to 11 is implemented.

24. An electronic device, characterized in that: include: processor; as well as a memory for storing executable instructions of the processor; The processor is configured to execute the information list recall method according to any one of claims 1 to 11 by executing the executable instructions.

Images