Tag-based search method, device, server and storage medium

Through the tag-based search method, the tags entered by the user are analyzed and combined with user behavior data, the problem of mismatch between user attributes and keywords in the prior art is solved, and higher search accuracy and accurate matching of user information is achieved.

CN111695031BActive Publication Date: 2024-12-06CHINA PING AN LIFE INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010347401.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-04-28
Publication Date
2024-12-06
Estimated Expiration
2040-04-28

AI Technical Summary

Technical Problem

In existing search technology, the keywords entered by the user do not match the user's attributes, resulting in low search accuracy and inability to accurately find the target user.

Method used

A tag-based search method is adopted to analyze the tags entered by the user through the first search engine, generate identifiable search conditions, and use the second search engine to match the user database, combining user behavior data and tag identification model to improve search accuracy.

Benefits of technology

Improve the accuracy of user searches, ensure that users can accurately find matching search results, and reduce search failures caused by mismatch between keywords and attributes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111695031B_ABST
    Figure CN111695031B_ABST
Patent Text Reader

Abstract

The present application relates to blockchain and artificial intelligence. The embodiment of the present application provides a tag-based search method, device, server and storage medium, the method comprising: receiving a user search request sent by a terminal, the user search request carrying a first set of tags for user search; parsing a second set of tags obtained according to the first set of tags by a first search engine to obtain a parsing result, the parsing result including a search condition that can be recognized by a second search engine; querying at least one user information matching the parsing result from a first database according to a tag set of each user among multiple users by the second search engine, the tag set of each user including at least one tag of the user, the first database including user information of each user among the multiple users; obtaining a user search result according to the at least one user information, and returning the user search result to the terminal. The present application can improve the accuracy of user search.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a tag-based search method, device, server and storage medium. Background Art

[0002] With the rapid development of Internet technology, mobile Internet technology and other technologies, the data generated by various data platforms has shown explosive growth. Massive data are distributed in every corner of digital life. How to effectively search for data has become a hot topic of research.

[0003] Generally speaking, during the user search process, a keyword about the user's attributes can be entered into the search engine. At this time, the search results based on the keyword may show many users with this keyword, which requires further investigation. In addition, during the search for the user, there is a situation where the entered keyword does not match the user's attributes, which results in the user not being found. Therefore, the above method results in a low accuracy of user search. Summary of the invention

[0004] The embodiments of the present application provide a tag-based search method, device, server, and storage medium, which can improve user search accuracy.

[0005] In a first aspect, an embodiment of the present application provides a tag-based search method, comprising:

[0006] Receiving a user search request sent by a terminal, the user search request carrying a first group of tags for user search, the first group of tags including one or more tags;

[0007] Parsing the second group of tags obtained according to the first group of tags by the first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine;

[0008] querying, by the second search engine, at least one piece of user information matching the parsing result from a first database according to a tag set of each user among a plurality of users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the plurality of users;

[0009] A user search result is obtained according to the at least one user information, and the user search result is returned to the terminal.

[0010] Optionally, before receiving the user search request sent by the terminal, the method further includes:

[0011] Acquire a correspondence table between user attributes and tags from a second database, wherein the correspondence table between user attributes and tags includes a plurality of user attributes and at least one tag corresponding to each of the plurality of user attributes;

[0012] Obtain at least one user attribute corresponding to each of the multiple users;

[0013] According to at least one user attribute corresponding to each of the multiple users and a correspondence table between the user attributes and tags, a tag adding process is performed on each of the at least one user attribute corresponding to each user to obtain a tag set for each user.

[0014] Optionally, the method further comprises:

[0015] Obtaining behavior data of each of the multiple users;

[0016] Performing feature analysis on the behavior data of each of the multiple users to obtain a behavior feature of each of the multiple users;

[0017] Inputting the behavior feature of each user among the multiple users into a label recognition model for label recognition, and obtaining a label corresponding to the behavior feature of each user among the multiple users; the label recognition model is trained according to a preset training sample set, and the training sample set includes at least one feature parameter corresponding to each behavior feature among the multiple behavior features;

[0018] Add a tag corresponding to the behavior feature of each user among the multiple users to the tag set of the user.

[0019] Optionally, before inputting the behavior feature of each of the multiple users into a label recognition model to perform label recognition and obtaining a label corresponding to the behavior feature of each of the multiple users, the method further includes:

[0020] According to the training sample set, a function for drawing a scatter plot matrix is ​​called to construct a scatter plot matrix; the scatter plot matrix includes distribution data of each behavior feature in the multiple behavior features, and association relationship data between a first behavior feature and a second behavior feature in the multiple behavior features; the first behavior feature is any behavior feature in the multiple behavior features;

[0021] Determining a target feature set from the plurality of behavioral features according to the scatter plot matrix;

[0022] The preset model is trained using at least one feature parameter corresponding to each behavior feature in the target feature set included in the training sample set to obtain the trained preset model, and a label recognition model is obtained based on the trained model.

[0023] Optionally, determining a target feature set from the plurality of behavior features according to the scatter plot matrix includes:

[0024] Determining a feature set from the plurality of behavior features according to the scatter plot matrix, the feature set comprising a first number of behavior features;

[0025] A target feature set is determined from the feature set using a feature selection algorithm, where the target feature set includes a second number of behavior features, where the second number is less than or equal to the first number.

[0026] Optionally, there are multiple preset models, there are multiple trained preset models, and obtaining a second preset model according to the trained preset models includes:

[0027] Draw a PR curve of each of the trained preset models in the plurality of trained preset models by using a PR curve drawing tool;

[0028] According to the PR curve of each trained preset model, an optimal trained preset model is determined from the multiple trained preset models as the second preset model.

[0029] Optionally, after receiving the user search request sent by the terminal, parsing the second group of tags obtained according to the first group of tags by the first search engine, and before obtaining the parsing result, the method further includes:

[0030] Determine at least one tag combination result corresponding to the first group of tags according to the tag hierarchical relationship; each tag combination result in the at least one tag combination result is different;

[0031] Sending the at least one label combination result to the terminal;

[0032] When a confirmation instruction for a target tag combination result in the at least one tag combination result sent by the terminal is received, the target tag combination result is determined as a second group of tags.

[0033] In a second aspect, an embodiment of the present application provides a tag-based search device, including:

[0034] A communication module, configured to receive a user search request sent by a terminal, wherein the user search request carries a first group of tags for user search, wherein the first group of tags includes one or more tags;

[0035] A processing module, configured to parse the second set of tags obtained according to the first set of tags through a first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine;

[0036] The processing module is further configured to query, through the second search engine, at least one piece of user information matching the parsing result from the first database according to a tag set of each user among the multiple users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the multiple users;

[0037] The processing module is further configured to obtain a user search result according to the at least one user information, and return the user search result to the terminal through the communication module.

[0038] In a third aspect, an embodiment of the present application provides a server comprising a processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected, wherein the memory is used to store a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the method described in the first aspect.

[0039] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method described in the first aspect.

[0040] To summarize, the server can receive a user search request sent by a terminal, the user search request carries a first set of tags for user search, and parses a second set of tags obtained based on the first set of tags through a first search engine to obtain a parsing result, and the parsing result includes a search condition that can be recognized by a second search engine; the server can query at least one user information matching the parsing result from a first database based on a tag set of each user among multiple users through the second search engine, and obtain a user search result based on the at least one user information to return to the terminal. Compared with the prior art where the user cannot be found due to the mismatch between the input keywords and the user attributes, the present application adopts the first search engine, the second search engine and the tag set to improve the user search accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 It is a flowchart of a tag-based search method provided in an embodiment of the present application;

[0043] Figure 2 is a flowchart of another tag-based search method provided in an embodiment of the present application;

[0044] Figure 3 is a schematic diagram of a network architecture of a tag-based search system provided in an embodiment of the present application;

[0045] Figure 4 is a structural diagram of a tag-based search device provided in an embodiment of the present application;

[0046] Figure 5 It is a structural diagram of a server provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0048] See also Figure 1 , is a flowchart of a tag-based search method provided in an embodiment of the present application. The method can be applied to a server, which can be a service device or a service device cluster composed of multiple service devices. Specifically, the method can include the following steps:

[0049] S101. Receive a user search request sent by a terminal, where the user search request carries a first group of tags for user search, and the first group of tags includes one or more tags.

[0050] The terminal can send a user search request to the server, and the server can receive the user search request sent by the terminal. The user search request can carry a first group of tags for user search, and the first group of tags includes one or more tags. The tag here can be represented by a tag identifier, which can be a name or a serial number. In one embodiment, the tag can also be represented by a tag identifier and a tag description information corresponding to the tag. In one embodiment, the tag can also be represented by a tag identifier, a tag description information corresponding to the tag, and a tag type corresponding to the tag. The tag type can include at least one of the following: single choice, multiple choice, button, drop-down box, text input box. The form of expression of the tag can be determined according to actual business needs, and the embodiments of the present application do not limit this.

[0051] In an application scenario, a terminal user may use a terminal to open a search page and input a first set of tags based on an input box included in the search page. After inputting the first set of tags, the terminal user may click a search button included in the search page. In response to the click operation on the search button, the terminal may send a user search request to a server, and the user search request carries the first set of tags. The server may receive the user search request sent by the terminal. In an application scenario, the search page may be a search page for user search.

[0052] In another application scenario, a terminal user can use a terminal to open a search page, and input elements for constructing a first set of tags based on an input box included in the search page. After inputting the elements for constructing the first set of tags, the terminal user can click a search button included in the search page. The terminal can respond to the click operation of the search button, construct the first set of tags according to the elements for constructing the first set of tags, and send a user search request to the server, the user search request carrying the first set of tags. The server can receive the user search request sent by the terminal.

[0053] In one embodiment, when the first group of tags includes multiple tags, the first group of tags may also include logical conjunctions for connecting the multiple tags. For example, the first group of tags may be tag 1, tag 2, and tag 3. The logical conjunctions may also be called logical conjunctions. In one embodiment, in addition to the logical conjunctions, the first group of tags may also include brackets. For example, the first group of tags may be (tag 1 and tag 2) or tag 3.

[0054] S102: parsing the second group of tags obtained according to the first group of tags through a first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine.

[0055] Since the first set of tags may be relatively complex, the server parses the second set of tags obtained from the first set of tags through the first search engine to obtain the parsing results, which can improve the search efficiency and search accuracy of the second search engine. The second set of tags may be the first set of tags. In one embodiment, the second set of tags may not be the first set of tags. The search condition may be understood as a search rule.

[0056] S103. Query at least one user information matching the parsing result from the first database through the second search engine according to the tag set of each user among the multiple users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the multiple users.

[0057] In the embodiment of the present application, in order to further ensure the privacy and security of the above-mentioned user information, the above-mentioned user information can also be stored in a node of a blockchain, that is, the database of this solution can be the database of any node on the blockchain. The blockchain referred to in the present invention is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, which is a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. The blockchain can include the underlying blockchain platform, the platform product service layer, and the application service layer.

[0058] In an embodiment of the present application, the server may include a tag set for each of the multiple users. In one embodiment, the tag set for each of the multiple users included in the server may be obtained from the first database. Or, the second search engine may include a tag set for each of the multiple users. In one embodiment, the tag set for each of the multiple users included in the second search engine may be obtained from the first database.

[0059] In one embodiment, the process of the server querying at least one user information matching the parsing result from the first database according to the tag set of each user among the multiple users through the second search engine can be as follows: the server can determine the user matching the parsing result according to the tag set of each user among the multiple users through the second search engine, and query the user information of the matching user from the first database. For example, the multiple users include user 1, user 2, and user 3, the tag set of user 1 includes tag 1, tag 2, and tag 3, the tag set of user 2 includes tag 1 and tag 3, the tag set of user 3 includes tag 4 and tag 5, and the first group of tags is: tag 1, tag 2, and tag 3. The server can determine that the user matching the parsing result is user 1 according to the tag set of each user among the multiple users through the second search engine, and then query the user information of user 1 from the first database.

[0060] In one embodiment, the server can obtain the identification information of the user matching the parsing result according to the tag set of each user among the multiple users through the second search engine, and query the user information of the matching user from the first database according to the identification information of the matching user. The identification information includes but is not limited to information such as name and login number for uniquely identifying the user.

[0061] In one embodiment, the server may determine the degree of match between the user and the parsed result based on the tag set of each user among the multiple users by using a second search engine using a fuzzy search method, and query the user information of the user whose matching degree is greater than a preset matching degree from the first database. In one embodiment, the process may be performed after the user matching the parsed result cannot be determined by the second search engine based on the tag set of each user among the multiple users. In one embodiment, the server may query at least one user information matching the parsed result from the first database based on the tag set of each user among the multiple users by using the second search engine.

[0062] In one embodiment, the first database may include a tag set of each user among the multiple users, and the server may query at least one user information matching the parsing result from the first database through the second search engine. For example, the server may send a query instruction to the first database; the first database may respond to the query instruction and query at least one user information matching the parsing result according to the tag set of each user among the multiple users.

[0063] S104: Obtain a user search result according to the at least one user information, and return the user search result to the terminal.

[0064] In the embodiment of the present application, the server may obtain a user search result based on the at least one user information, and return the user search result to the terminal, and the terminal may display the user search result.

[0065] In one embodiment, the process of the server obtaining the user search result according to the at least one user information may generate a user search result including the at least one user information for the server.

[0066] In one embodiment, the process of the server obtaining the user search results according to the at least one user information may include the server performing data processing on the at least one user information, obtaining the processing result, and generating the user search results including the processing result. For example, the data processing may include information screening or sorting according to the matching degree between the user and the first set of tag sets.

[0067] visible, Figure 1 In the illustrated embodiment, the server can receive a user search request sent by a terminal, the user search request carries a first set of tags for user search, and parses a second set of tags obtained based on the first set of tags through a first search engine to obtain a parsing result, the parsing result including search conditions that can be recognized by the second search engine; the server can query at least one user information matching the parsing result from the first database based on the tag set of each user among multiple users through the second search engine, obtain a user search result based on the at least one user information, and return the user search result to the terminal, thereby improving the accuracy of the user search.

[0068] See also Figure 2 , is a flow chart of another tag-based search method provided in an embodiment of the present application. The method can be applied to a server, which can be a service device or a service device cluster composed of multiple service devices. Specifically, the method can include the following steps:

[0069] S201. Receive a user search request sent by a terminal, where the user search request carries a first group of tags for user search, and the first group of tags includes one or more tags.

[0070] Among them, step S201 can be seen in Figure 1 Step S101 in the embodiment is not described in detail in the embodiment of the present application.

[0071] S202: Determine at least one tag combination result corresponding to the first group of tags according to the tag hierarchical relationship.

[0072] In an embodiment of the present application, the server may determine at least one tag combination result corresponding to the first group of tags, such as determining at least one tag combination result corresponding to the first group of tags according to the tag hierarchical relationship in step S203. Among them, the tag hierarchical relationship may refer to the hierarchical relationship between tags, such as the hierarchical relationship between a first-level tag and each sub-tag under the first-level tag (such as a first-level sub-tag and a second-level sub-tag). In one embodiment, the tag hierarchical relationship may be recorded in a correspondence table between user attributes and tags or in a tag library. Among them, the tag library may include a large number of tags, such as tags that may include the correspondence table between the user attributes and tags. In one embodiment, the tag library may be recorded in a correspondence table between user attributes and tags or may also be recorded in a database table, and the embodiment of the present application does not limit this.

[0073] Among them, one or more label combination results mentioned in step S203 may include a parent label or a child label of any one or more labels in the first group of labels. In one embodiment, in one embodiment, when the label combination result includes a parent label (or child label) of any one or more labels in the first group of labels, the label combination result may not include the any one or more labels, thereby preventing duplication of similar labels. In one embodiment, when the label combination result includes a parent label (or child label) of any one or more labels, the label combination result may also include other labels in the first group of labels except the any one or more labels.

[0074] Correspondingly, the process of the server determining at least one tag combination result corresponding to the first group of tags according to the tag hierarchy relationship can be that the server queries the parent tag (or child tag) corresponding to the target tag in the first group of tags according to the tag hierarchy relationship, and the target tag is any one or more tags in the first group of tags; the server obtains at least one tag combination result according to the parent tag (or child tag). In one embodiment, the server can combine the target tags to obtain at least one combination result. Or, the server can combine the parent tag (or child tag) corresponding to the target tag and other tags in the first group of tags except the target tag to obtain at least one tag combination result.

[0075] In one embodiment, each of the at least one tag combination result may include one or more tags in the first group of tags. In one embodiment, each of the at least one combination result may also include tags whose similarity with one or more tags in the first group of tags is greater than or equal to a preset similarity.

[0076] In one embodiment, the process of the server determining at least one tag combination result corresponding to the first group of tags may also be that the server combines the first group of tags to obtain at least one combination result.

[0077] In one embodiment, the process of the server determining at least one tag combination result corresponding to the first group of tags may also be that the server determines a second tag whose similarity between the first tags in the first group of tags is greater than or equal to a preset similarity, and obtains at least one tag combination result based on the second tag. Wherein, the first tag is any one or more tags in the first group of tags. In one embodiment, the process of the server obtaining at least one tag combination result based on the second tag may be that the server combines the second tags to obtain at least one tag combination result; or, the server combines the second tag and other tags in the first group of tags except the first tag to obtain at least one tag combination result.

[0078] In one embodiment, the server may also determine the attributes corresponding to each tag in the first group of tags according to the correspondence table between user attributes and tags, and determine the associated attributes of the attributes corresponding to each tag, and then combine at least one tag under the associated attributes of the attributes corresponding to each tag to obtain at least one tag combination result. For example, the first group of tags is: tag 1 and tag 2, tag 1 corresponds to attribute 1, and tag 2 corresponds to attribute 2. The server may combine at least one tag under the associated attributes of attribute 1 with at least one tag under the associated attributes of attribute 2 to obtain at least one tag combination result; or, the server may also combine at least one tag under the associated attributes of attribute 1 (such as between tags in a parallel relationship), and combine at least one tag under the associated attributes of attribute 2 (such as between tags in a parallel relationship) to obtain at least one tag combination result.

[0079] In one embodiment, the server may also perform splicing processing on each of the at least one label combination result to obtain a plurality of label splicing results corresponding to each label combination result. The splicing process may include a process of adding logical conjunctions to each label combination result. For example, the first group of labels are: label 1 and label 2. The at least one label combination result corresponding to the first group of labels may include label combination result 1: label 1 label 2. After performing splicing processing on label combination result 1, the server may obtain the following label splicing results: label 1 and label 2, label 1 or label 2, non-label 1 and non-label 2. In one embodiment, the splicing process may also include a process of adding brackets to each. The server may send the plurality of label splicing results corresponding to each label combination result to the terminal.

[0080] In one embodiment, the server may also perform splicing processing on each of the at least one label combination result according to the logical connection relationship indicated by the first group of labels, and obtain multiple label splicing results corresponding to each label combination result. In this way, a label combination result consistent with the logical connection relationship indicated by the first group of labels can be obtained, and the purpose of accurately pushing the label combination result to the user can be achieved.

[0081] S203: Send the at least one label combination result to the terminal.

[0082] The server may send the at least one label combination result to the terminal or the label splicing result of each label combination result in the at least one label combination result to the terminal. In one embodiment, the server may send the at least one label combination result and the multiple label splicing results corresponding to each label combination result in the at least one label combination result to the terminal.

[0083] In one embodiment, the server may return search prompt information to the terminal, where the search prompt information carries the at least one tag combination result and / or multiple tag splicing results corresponding to each tag combination result in the at least one tag combination result.

[0084] In one embodiment, the server may further return a page including the at least one tag combination result and / or a plurality of tag splicing results corresponding to each tag combination result in the at least one tag combination result to the terminal.

[0085] In one embodiment, the server may also convert the tag combination result and / or the tag splicing result into a preset format, such as JSON format, and return it to the terminal, thereby improving the data parsing efficiency of the terminal.

[0086] S204: When receiving a confirmation instruction for a target tag combination result in the at least one tag combination result sent by the terminal, determine the target tag combination result as a second group of tags.

[0087] The terminal may send a confirmation instruction for the target tag combination result in the at least one tag combination result to the server, and the server may determine the target tag combination result as the second set of tags when receiving the confirmation instruction for the target tag combination result in the at least one tag combination result sent by the terminal. The target tag combination result may be a tag combination result selected from at least one tag combination result, or the target tag set may also be a tag combination result selected from at least one tag combination result and a plurality of tag splicing results corresponding to each tag combination result in the at least one tag combination result.

[0088] In one embodiment, the terminal may also send a confirmation instruction for the target tag splicing result to the server, and the server may determine the target tag splicing result as the second set of tags when receiving the confirmation instruction for the target tag splicing result sent by the terminal. The target tag splicing result is a tag splicing result selected from a plurality of tag splicing results corresponding to each tag combination result in at least one tag combination result, or, or the target tag set may also be a tag splicing result selected from at least one tag combination result and a plurality of tag splicing results corresponding to each tag combination result in the at least one tag combination result.

[0089] S205: parse the second group of tags obtained according to the first group of tags through the first search engine to obtain a parsing result, where the parsing result includes a search condition that can be recognized by the second search engine.

[0090] S206. Query at least one piece of user information matching the parsing result from the first database through the second search engine according to the tag set of each user among the multiple users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the multiple users.

[0091] S207: Obtain a user search result according to the at least one user information, and return the user search result to the terminal.

[0092] Among them, steps S205-S207 can refer to Figure 1 Steps S102-S104 in the embodiment are not described in detail in the embodiment of the present application.

[0093] In one embodiment, the server may obtain a correspondence table between user attributes and tags from the second database, wherein the correspondence table between user attributes and tags includes multiple user attributes and at least one tag corresponding to each of the multiple user attributes. The tag may be a first-level tag or a sub-tag under a first-level tag, such as a first-level sub-tag or a second-level sub-tag. The server obtains at least one user attribute corresponding to each of the multiple users, and performs tag addition processing on each of the at least one user attribute corresponding to each of the multiple users and the user attribute-tag correspondence table to obtain a tag set for each user.

[0094] In one embodiment, the server may obtain a tag list and a user attribute list, wherein the tag list includes multiple tags and the user attribute list includes multiple user attributes. The server may establish a correspondence table including multiple user attributes and multiple tags corresponding to each of the multiple user attributes according to the multiple tags included in the tag list and the multiple user attributes included in the user attribute list.

[0095] In one embodiment, the server may obtain the behavior data of each user among the multiple users, and obtain the tag set of the user according to the behavior data of each user among the multiple users and the tag recognition model.

[0096] In one embodiment, the process in which the server obtains the behavior data of each user among the multiple users, and obtains the tag set of the user based on the behavior data of each user among the multiple users and the tag recognition model can also be as follows: the server obtains the behavior data of each user among the multiple users, and performs feature analysis on the behavior data of each user among the multiple users to obtain the behavior characteristics of each user among the multiple users; the server inputs the behavior characteristics of each user among the multiple users into the tag recognition model for tag recognition, and obtains the tag corresponding to the behavior characteristics of each user among the multiple users; the tag recognition model is trained according to a preset training sample set, and the training sample set includes at least one feature parameter corresponding to each behavior characteristic among the multiple behavior characteristics; the server adds the tag corresponding to the behavior characteristic of each user among the multiple users to the tag set of the user.

[0097] In one embodiment, before the server inputs the behavior characteristics of each user in the multiple users into the label recognition model for label recognition and obtains the label corresponding to the behavior characteristics of each user in the multiple users, it can call the function of drawing a scatter plot matrix to construct a scatter plot matrix according to the training sample set; the scatter plot matrix includes the distribution data of each behavior characteristic in the multiple behavior characteristics, and the association relationship data between the first behavior characteristic and the second behavior characteristic in the multiple behavior characteristics. The distribution data here includes but is not limited to being displayed in the form of a histogram, a density map, such as a kernel density map, etc. The association relationship data includes but is not limited to being displayed in the form of a scatter plot. The first behavior characteristic is any behavior characteristic in the multiple behavior characteristics. In one embodiment, the scatter plot matrix may also include the association relationship data between the first behavior characteristic and each label in the preset multiple labels. The server determines the target feature set from the multiple behavior characteristics according to the scatter plot matrix; the server trains the preset model using at least one feature parameter corresponding to each behavior characteristic in the target feature set included in the training sample set to obtain the trained preset model, and obtains the label recognition model according to the trained model.

[0098] In one embodiment, the process of the server determining the target feature set from the multiple behavior features according to the scatter plot matrix may be that the server determines a feature set from the multiple behavior features as the target feature set according to the scatter plot matrix, and the feature set includes a first number of behavior features. In one embodiment, the server may calculate the correlation coefficient between each behavior feature in the multiple behavior features and other behavior features in the multiple behavior features except the behavior feature, and determine the feature set from the multiple behavior features according to the correlation coefficient and the scatter plot matrix. Alternatively, the server may also calculate the correlation coefficient between each behavior feature in the multiple behavior features and each tag in the preset multiple tags, and determine the feature set from the multiple behavior features according to the correlation coefficient and the scatter plot matrix.

[0099] In one embodiment, the process of the server determining the target feature set from the multiple behavioral features according to the scatter plot matrix can also be that the server determines the feature set from the multiple behavioral features according to the scatter plot matrix; the server uses a feature selection algorithm to determine the target feature set from the feature set, and the target feature set includes a second number of behavioral features, and the second number is less than or equal to the first number. For example, the feature selection algorithm can be a scoring-based method, through which each behavioral feature in the feature set can be scored, and the target feature set is determined from the feature set according to the scoring result.

[0100] In one embodiment, this scheme utilizes the model algorithm technology in artificial intelligence, there are multiple preset models, there are multiple trained preset models, and the process of obtaining the second preset model according to the trained preset model can be: the server draws the PR curve of each trained preset model in multiple trained preset models through a PR curve drawing tool; the server determines the optimal trained preset model from the multiple trained preset models as the second preset model according to the PR curve of each trained preset model. Among them, the PR curve refers to the precision-recall curve, or refers to the precision-recall curve. In one embodiment, the PR curve can also be replaced by other model parameter curves, such as the Receiver Operating Characteristic (ROC) curve.

[0101] In one embodiment, the server obtains the behavior data of each user among the multiple users, and obtains the tag set of the user based on the behavior data of each user among the multiple users and the tag recognition model. The process can also be as follows: the server can obtain the behavior data of each user among the multiple users, input the behavior data of each user among the multiple users into the specified tag recognition model for tag recognition, and obtain the tag corresponding to the behavior data of each user among the multiple users; the server adds the tag corresponding to the behavior data of each user among the multiple users to the tag set of the user.

[0102] visible, Figure 2 In the illustrated embodiment, after receiving a user search request sent by a terminal, the server can determine at least one tag combination result corresponding to the first group of tags based on a correspondence table between user attributes and tags, and send the at least one tag combination result to the terminal; when the server receives a confirmation instruction sent by the terminal for a target tag combination result in the at least one tag combination result, the server can determine the target tag combination result as a second group of tags, thereby obtaining user search results based on the second group of tags and returning them to the terminal. This process can improve the accuracy of user searches.

[0103] See also Figure 3 , is a schematic diagram of a network architecture of a tag-based search system provided in an embodiment of the present application. Figure 3 The search system shown may include a terminal 10 and a server 20. Specifically:

[0104] The terminal 10 may send a user search request to the server 20. The server 20 may obtain the user search request by executing step S101, and may obtain the user search result by executing steps S102-S104 to return it to the terminal 10, thereby improving the accuracy of the user search process.

[0105] See also Figure 4 , is a schematic diagram of a structure of a tag-based search device provided in an embodiment of the present application. The device can be applied to a server. Specifically, the device may include:

[0106] The communication module 401 is configured to receive a user search request sent by a terminal, wherein the user search request carries a first group of tags for user search, and the first group of tags includes one or more tags.

[0107] The processing module 402 is used to parse the second group of tags obtained according to the first group of tags through the first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine.

[0108] The processing module 402 is also used to query at least one user information matching the parsing result from the first database through the second search engine according to the tag set of each user among the multiple users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the multiple users.

[0109] The processing module 402 is further configured to obtain a user search result according to the at least one user information, and return the user search result to the terminal through the communication module 401 .

[0110] In an optional implementation, the processing module 402 is further used to obtain a correspondence table between user attributes and tags from a second database before receiving a user search request sent by a terminal, wherein the correspondence table between user attributes and tags includes multiple user attributes and at least one tag corresponding to each of the multiple user attributes; obtain at least one user attribute corresponding to each of the multiple users; and perform tag adding processing on each of the at least one user attribute corresponding to each user according to the at least one user attribute corresponding to each of the multiple users and the correspondence table between the user attributes and tags to obtain a tag set for each user.

[0111] In an optional embodiment, the processing module 402 is further used to obtain behavior data of each user among the multiple users; perform feature analysis on the behavior data of each user among the multiple users to obtain behavior features of each user among the multiple users; input the behavior features of each user among the multiple users into a label recognition model for label recognition to obtain a label corresponding to the behavior features of each user among the multiple users; the label recognition model is trained according to a preset training sample set, and the training sample set includes at least one feature parameter corresponding to each behavior feature among the multiple behavior features; and add the label corresponding to the behavior features of each user among the multiple users to the user's label set.

[0112] In an optional embodiment, the processing module 402 is also used to call a function of drawing a scatter plot matrix to construct a scatter plot matrix according to the training sample set before inputting the behavioral characteristics of each user among the multiple users into the label recognition model for label recognition and obtaining the label corresponding to the behavioral characteristics of each user among the multiple users; the scatter plot matrix includes distribution data of each behavioral feature among the multiple behavioral features, and association relationship data between a first behavioral feature and a second behavioral feature among the multiple behavioral features; the first behavioral feature is any behavioral feature among the multiple behavioral features; a target feature set is determined from the multiple behavioral features according to the scatter plot matrix; a preset model is trained using at least one feature parameter corresponding to each behavioral feature in the target feature set included in the training sample set to obtain the trained preset model, and a label recognition model is obtained based on the trained model.

[0113] In an optional embodiment, the processing module 402 determines a target feature set from the multiple behavioral features according to the scatter plot matrix, specifically determines a feature set from the multiple behavioral features according to the scatter plot matrix, and the feature set includes a first number of behavioral features; uses a feature selection algorithm to determine a target feature set from the feature set, and the target feature set includes a second number of behavioral features, and the second number is less than or equal to the first number.

[0114] In an optional embodiment, there are multiple preset models. There are multiple preset models trained by the processing module 402, and the second preset model is obtained according to the trained preset models, specifically by drawing the PR curve of each trained preset model in the multiple trained preset models through a PR curve drawing tool; according to the PR curve of each trained preset model, the optimal trained preset model is determined from the multiple trained preset models as the second preset model.

[0115] In an optional embodiment, the processing module 402 is also used to parse the second group of tags obtained according to the first group of tags through the first search engine after receiving the user search request sent by the terminal through the communication module 401, and before obtaining the parsing result, determine at least one tag combination result corresponding to the first group of tags according to the tag hierarchy relationship; each tag combination result in the at least one tag combination result is different; send the at least one tag combination result to the terminal through the communication module 401; when a confirmation instruction for the target tag combination result in the at least one tag combination result sent by the terminal is received through the communication module 401, the target tag combination result is determined as the second group of tags.

[0116] visible, Figure 4 In the illustrated embodiment, the search device can receive a user search request sent by a terminal, the user search request carries a first set of tags for user search, and parses a second set of tags obtained based on the first set of tags through a first search engine to obtain a parsing result, the parsing result including search conditions that can be recognized by the second search engine; the search device can query at least one user information matching the parsing result from the first database based on the tag set of each user among multiple users through the second search engine, obtain a user search result based on the at least one user information, and return the user search result to the terminal, thereby improving the accuracy of the user search.

[0117] See also Figure 5 , is a schematic diagram of the structure of a server provided in an embodiment of the present application. The server described in this embodiment may include: a processor 1000, an input device 2000, an output device 3000, and a memory 4000. The processor 1000, the input device 2000, the output device 3000, and the memory 4000 may be connected via a bus or other means. The input device 2000 and the output device 3000 may be standard wired or wireless communication interfaces.

[0118] The processor 1000 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0119] The memory 4000 may be a high-speed RAM memory or a non-volatile memory, such as a disk memory. The memory 4000 is used to store a set of program codes, and the processor 1000, the input device 2000, and the output device 3000 may call the program codes stored in the memory 4000. Specifically:

[0120] The processor 1000 is configured to receive a user search request sent by a terminal through an input device 2000, wherein the user search request carries a first group of tags for user search, wherein the first group of tags includes one or more tags; parse a second group of tags obtained based on the first group of tags through a first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine; query at least one user information matching the parsing result from a first database based on a tag set of each user among a plurality of users through the second search engine, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the plurality of users; obtain a user search result based on the at least one user information, and return the user search result to the terminal through an output device 3000.

[0121] In one embodiment, the processor 1000 is further used to obtain a correspondence table between user attributes and tags from a second database before receiving a user search request sent by a terminal through an input device 2000, wherein the correspondence table between user attributes and tags includes multiple user attributes and at least one tag corresponding to each of the multiple user attributes; obtain at least one user attribute corresponding to each of the multiple users; and perform tag adding processing on each of the at least one user attribute corresponding to each user according to the at least one user attribute corresponding to each of the multiple users and the correspondence table between the user attributes and tags to obtain a tag set for each user.

[0122] In one embodiment, the processor 1000 is further used to obtain behavior data of each of the multiple users; perform feature analysis on the behavior data of each of the multiple users to obtain behavior features of each of the multiple users; input the behavior features of each of the multiple users into a label recognition model for label recognition to obtain a label corresponding to the behavior features of each of the multiple users; the label recognition model is trained based on a preset training sample set, and the training sample set includes at least one feature parameter corresponding to each of the multiple behavior features; and add the label corresponding to the behavior features of each of the multiple users to the user's label set.

[0123] In one embodiment, the processor 1000 is also used to call a function of drawing a scatter plot matrix to construct a scatter plot matrix according to the training sample set before inputting the behavior characteristics of each user among the multiple users into the label recognition model for label recognition and obtaining the label corresponding to the behavior characteristics of each user among the multiple users; the scatter plot matrix includes distribution data of each behavior characteristic among the multiple behavior characteristics, and association relationship data between a first behavior characteristic and a second behavior characteristic among the multiple behavior characteristics; the first behavior characteristic is any behavior characteristic among the multiple behavior characteristics; a target feature set is determined from the multiple behavior characteristics according to the scatter plot matrix; a preset model is trained using at least one feature parameter corresponding to each behavior characteristic in the target feature set included in the training sample set to obtain the trained preset model, and a label recognition model is obtained according to the trained model.

[0124] In one embodiment, the processor 1000 determines a target feature set from the multiple behavioral features based on the scatter plot matrix, specifically determines a feature set from the multiple behavioral features based on the scatter plot matrix, and the feature set includes a first number of behavioral features; uses a feature selection algorithm to determine a target feature set from the feature set, and the target feature set includes a second number of behavioral features, and the second number is less than or equal to the first number.

[0125] In one embodiment, there are multiple preset models, and there are multiple preset models trained by the processor 1000. The second preset model is obtained according to the trained preset models, specifically by drawing the PR curve of each trained preset model in the multiple trained preset models through a PR curve drawing tool; according to the PR curve of each trained preset model, the optimal trained preset model is determined from the multiple trained preset models as the second preset model.

[0126] In one embodiment, the processor 1000 is further used to parse the second group of tags obtained according to the first group of tags through the first search engine after receiving a user search request sent by the terminal through the input device 2000, and before obtaining the parsed result, determine at least one tag combination result corresponding to the first group of tags according to the tag hierarchy relationship; each tag combination result in the at least one tag combination result is different; send the at least one tag combination result to the terminal through the output device 3000; and when a confirmation instruction for a target tag combination result in the at least one tag combination result sent by the terminal is received through the input device 2000, determine the target tag combination result as the second group of tags.

[0127] In a specific implementation, the processor 1000, the input device 2000, and the output device 3000 described in the embodiments of the present application may execute Figure 1 and Figure 2 The implementation method described in the embodiment can also execute the implementation method described in the embodiment of the present application, which will not be repeated here.

[0128] Each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of sampling hardware or in the form of sampling software functional unit.

[0129] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The computer-readable storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0130] What is disclosed above is only a preferred embodiment of the present application, and it certainly cannot be used to limit the scope of rights of the present application. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiment and equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A tag-based search method, It is characterized in that include: Receiving a user search request sent by a terminal, the user search request carrying a first group of tags for user search, the first group of tags including one or more tags; Parsing the second group of tags obtained according to the first group of tags by the first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine; querying, by the second search engine, at least one piece of user information matching the parsing result from a first database according to a tag set of each user among a plurality of users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the plurality of users; Obtaining a user search result according to the at least one user information, and returning the user search result to the terminal; The method further comprises: Obtaining behavior data of each of the multiple users; Performing feature analysis on the behavior data of each of the multiple users to obtain a behavior feature of each of the multiple users; Inputting the behavior feature of each user among the multiple users into a label recognition model for label recognition, and obtaining a label corresponding to the behavior feature of each user among the multiple users; the label recognition model is trained according to a preset training sample set, and the training sample set includes at least one feature parameter corresponding to each behavior feature among the multiple behavior features; adding the label corresponding to the behavior feature of each user among the multiple users to the label set of the user; Before inputting the behavior characteristics of each user among the plurality of users into the tag recognition model for tag recognition, the method further includes: According to the training sample set, a function for drawing a scatter plot matrix is ​​called to construct a scatter plot matrix; the scatter plot matrix includes distribution data of each behavior feature in the multiple behavior features, and association relationship data between a first behavior feature and a second behavior feature in the multiple behavior features; the first behavior feature is any behavior feature in the multiple behavior features; Determining a target feature set from the plurality of behavioral features according to the scatter plot matrix; The preset model is trained using at least one feature parameter corresponding to each behavior feature in the target feature set included in the training sample set to obtain the trained preset model, and a label recognition model is obtained based on the trained model.

2. The method according to claim 1, It is characterized in that Before receiving the user search request sent by the terminal, the method further includes: Acquire a correspondence table between user attributes and tags from a second database, wherein the correspondence table between user attributes and tags includes a plurality of user attributes and at least one tag corresponding to each of the plurality of user attributes; Obtain at least one user attribute corresponding to each of the multiple users; According to at least one user attribute corresponding to each of the multiple users and a correspondence table between the user attributes and tags, a tag adding process is performed on each of the at least one user attribute corresponding to each user to obtain a tag set for each user.

3. The method according to claim 1, It is characterized in that Determining a target feature set from the plurality of behavior features according to the scatter plot matrix includes: Determining a feature set from the plurality of behavior features according to the scatter plot matrix, the feature set comprising a first number of behavior features; A target feature set is determined from the feature set using a feature selection algorithm, where the target feature set includes a second number of behavior features, where the second number is less than or equal to the first number.

4. The method according to claim 1, It is characterized in that There are multiple preset models, there are multiple trained preset models, and obtaining a second preset model according to the trained preset models includes: Draw a PR curve of each of the trained preset models in the plurality of trained preset models by using a PR curve drawing tool; According to the PR curve of each trained preset model, an optimal trained preset model is determined from the multiple trained preset models as the second preset model.

5. The method according to claim 1, It is characterized in that After receiving the user search request sent by the terminal, the method further includes: parsing the second group of tags obtained according to the first group of tags by the first search engine, and before obtaining the parsing result: Determine at least one tag combination result corresponding to the first group of tags according to the tag hierarchical relationship; each tag combination result in the at least one tag combination result is different; Sending the at least one label combination result to the terminal; When a confirmation instruction for a target tag combination result in the at least one tag combination result sent by the terminal is received, the target tag combination result is determined as a second group of tags.

6. A tag-based search device, It is characterized in that include: A communication module, configured to receive a user search request sent by a terminal, wherein the user search request carries a first group of tags for user search, wherein the first group of tags includes one or more tags; A processing module, configured to parse the second set of tags obtained according to the first set of tags through a first search engine to obtain a parsing result, wherein the parsing result includes a search condition that can be recognized by the second search engine; The processing module is further configured to query, through the second search engine, at least one piece of user information matching the parsing result from the first database according to a tag set of each user among the multiple users, wherein the tag set of each user includes at least one tag of the user, and the first database includes user information of each user among the multiple users; The processing module is further configured to obtain a user search result according to the at least one user information, and return the user search result to the terminal through the communication module; The processing module is further used to obtain behavior data of each user among the multiple users; Performing feature analysis on the behavior data of each of the multiple users to obtain a behavior feature of each of the multiple users; Inputting the behavior feature of each user among the multiple users into a label recognition model to perform label recognition, and obtaining a label corresponding to the behavior feature of each user among the multiple users; The label recognition model is trained according to a preset training sample set, wherein the training sample set includes at least one feature parameter corresponding to each of a plurality of behavior features; a label corresponding to the behavior feature of each of the plurality of users is added to the label set of the user; The processing module is also used to call a function of drawing a scatter plot matrix to construct a scatter plot matrix according to the training sample set before inputting the behavioral characteristics of each user among the multiple users into the label recognition model for label recognition; the scatter plot matrix includes distribution data of each behavioral feature among the multiple behavioral features, and association relationship data between a first behavioral feature and a second behavioral feature among the multiple behavioral features; the first behavioral feature is any behavioral feature among the multiple behavioral features; a target feature set is determined from the multiple behavioral features according to the scatter plot matrix; a preset model is trained using at least one feature parameter corresponding to each behavioral feature in the target feature set included in the training sample set to obtain the trained preset model, and a label recognition model is obtained according to the trained model.

7. A server, It is characterized in that The method comprises a processor, an input device, an output device and a memory, wherein the processor, the input device, the output device and the memory are interconnected, wherein the memory is used to store a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 5.