Method and apparatus for determining data tag objects

Abstract

The application relates to a method and device for determining a data label object. The method comprises the following steps: obtaining target user object information of a target user object; obtaining a plurality of first data label objects based on the target user object information; determining a plurality of second data label objects with the highest association degree with the target user object, and obtaining a plurality of third data label objects similar to the plurality of second data label objects; obtaining a target user object feature vector of the target user object, and obtaining a plurality of fourth data label objects based on the target user object feature vector; and determining a target data label object to be recommended to the target user object based on the plurality of first data label objects, the plurality of third data label objects and the plurality of fourth data label objects. The method can improve the accuracy of data label object recommendation.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and apparatus for determining data tag objects. Background Technology

[0002] With the rapid development of information technology, data tagging objects are increasingly being managed uniformly by data management platforms to assist enterprises in making effective decisions. However, different users have different needs for data tagging objects, making it crucial to determine which data tagging objects to recommend to users a pressing issue.

[0003] Traditional data tag recommendation methods rely on directly analyzing the data tag's release date and historical access popularity to recommend newer, more popular data tag objects to all users. Therefore, it's clear that traditional data tag recommendation methods are not very accurate. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and apparatus for determining data-labeled objects that can improve the accuracy of data-labeled object recommendation, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for determining data tag objects. The method includes:

[0006] Obtain target user object information;

[0007] Based on the target user object information, query and obtain several similar user objects that are similar to the target user object, and obtain several first data tag objects that have the highest correlation with the several similar user objects;

[0008] Identify several second data tag objects that have the highest correlation with the target user object, and obtain several third data tag objects that are similar to the several second data tag objects;

[0009] Obtain the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vectors of the data tag objects corresponding to the target user object.

[0010] Based on the plurality of first data tag objects, the plurality of third data tag objects, and the plurality of fourth data tag objects, a target data tag object to be recommended to the target user object is determined.

[0011] In one embodiment, the query to retrieve a plurality of similar user objects that are similar to the target user object includes:

[0012] Obtain the target data tag vector corresponding to the target user object, and query several similar data tag vectors that are similar to the target data tag vector in the user object similarity matrix;

[0013] Determine the user objects corresponding to the plurality of similar data label vectors, wherein the plurality of similar user objects include the user objects corresponding to the plurality of similar data label vectors.

[0014] In one embodiment, the method for obtaining the user object similarity matrix includes:

[0015] Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object;

[0016] For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object;

[0017] The user object similarity matrix is ​​obtained based on the similarity of the data tag vectors corresponding to each user object.

[0018] In one embodiment, obtaining the plurality of first data tag objects with the highest correlation to the plurality of similar user objects includes:

[0019] In the data tag object rating table, the first data tag objects with the highest rating relationship with the several similar user objects are queried and determined. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0020] In one embodiment, the data label object rating table is established by means of:

[0021] Acquire batch behavior data, which includes: behavior data of each user object on each data tag object within a predetermined time range;

[0022] The batch behavior data is summarized using the user object as the primary key to obtain a summarized behavior data table;

[0023] Principal component analysis is performed on the summarized behavioral data table to obtain the data label object rating table, which includes the rating results of each user object for each data label object.

[0024] In one embodiment, obtaining a plurality of third data tag objects similar to the plurality of second data tag objects includes:

[0025] Obtain the user vector corresponding to each of the plurality of second data tag objects, and query the data tag object similarity matrix to obtain a plurality of similar user vectors that are similar to the user vectors;

[0026] Determine the data tag objects corresponding to the plurality of similar user vectors, and obtain the plurality of third data tag objects based on the data tag objects corresponding to the plurality of similar user vectors.

[0027] In one embodiment, the method for obtaining the data label object similarity matrix includes:

[0028] Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object;

[0029] For each of the data tag objects, the user object information of each user object of each of the data tag objects is vectorized to obtain the user vector corresponding to each of the data tag objects;

[0030] The similarity matrix of the data tag objects is obtained based on the similarity of the user vectors corresponding to each data tag object.

[0031] In one embodiment, the query for the plurality of data tag content vectors with the highest similarity to the feature vector of the target user object includes:

[0032] In the data tag content similarity matrix, several data tag content vectors that are similar to the feature vector of the target user object are queried and identified.

[0033] In one embodiment, the method for obtaining the data tag content similarity matrix includes:

[0034] Retrieve metadata information for each data tag object;

[0035] The metadata information of each data tag object is vectorized to obtain the word vector of each data tag object;

[0036] The content similarity matrix of the data tags is obtained based on the similarity of the word vectors of each data tag object.

[0037] Secondly, this application also provides an apparatus for determining data tag objects. The apparatus includes:

[0038] The information acquisition module is used to acquire target user object information.

[0039] The first object acquisition module is used to query and acquire several similar user objects that are similar to the target user object based on the target user object information, and acquire several first data tag objects that have the highest correlation with the several similar user objects;

[0040] The third object acquisition module is used to determine a number of second data tag objects that are most associated with the target user object, and to acquire a number of third data tag objects that are similar to the number of second data tag objects;

[0041] The fourth object acquisition module is used to acquire the target user object feature vector of the target user object, query several data tag content vectors with the highest similarity to the target user object feature vector, and obtain several fourth data tag objects corresponding to the several data tag content vectors. The target user object feature vector is determined based on the word vector of the data tag object corresponding to the target user object.

[0042] The recommendation object determination module is used to determine the target data tag object to be recommended to the target user object based on the plurality of first data tag objects, the plurality of third data tag objects, and the plurality of fourth data tag objects.

[0043] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the methods in any of the above embodiments.

[0044] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the methods in any of the above embodiments.

[0045] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the methods in any of the above embodiments.

[0046] The method and apparatus for determining data tag objects described above involve: acquiring target user object information of a target user object; acquiring several first data tag objects similar to the target user object based on the target user object information; determining several second data tag objects with the highest relevance to the target user object; acquiring several third data tag objects similar to the second data tag objects; acquiring the target user object feature vector of the target user object; acquiring several fourth data tag objects similar to the data tag content vector of the target user object based on the target user object feature vector; and determining the target data tag object to be recommended to the target user object based on the several first data tag objects, the several third data tag objects, and the several fourth data tag objects. Compared to the traditional technology where recommending data tag objects to users based on the popularity and novelty of data tag objects is not very accurate, this embodiment, through the target user object information, the several second data tag objects with the highest relevance to the target user object, and the data tag objects determined by the target user object feature vector of the target user object, can make more accurate recommendations for the target user object, solving the problem of inaccurate data tag object recommendations in the traditional technology. Attached Figure Description

[0047] Figure 1 This is an application environment diagram of the method for determining data tag objects provided in the embodiments of this application;

[0048] Figure 2 This is a flowchart illustrating the method for determining data tag objects provided in the embodiments of this application;

[0049] Figure 3 This is a flowchart illustrating the process of determining a first data tag object in one embodiment;

[0050] Figure 4 This is a flowchart illustrating how the user object similarity matrix is ​​obtained in one embodiment;

[0051] Figure 5 This is a flowchart illustrating how a data label object scoring table is created in one embodiment.

[0052] Figure 6 This is a flowchart illustrating the process of determining a third data tag object in one embodiment;

[0053] Figure 7 This is a flowchart illustrating how the data tag object similarity matrix is ​​obtained in one embodiment;

[0054] Figure 8 This is a flowchart illustrating the process of determining the fourth data tag object in one embodiment;

[0055] Figure 9This is a flowchart illustrating how the data tag content similarity matrix is ​​obtained in one embodiment;

[0056] Figure 10 This is a flowchart illustrating a data tag object recommendation method in one embodiment;

[0057] Figure 11 This is a structural block diagram of an apparatus for determining a data tag object provided in an embodiment of this application;

[0058] Figure 12 This is an internal structure diagram of a computer device provided in an embodiment of this application;

[0059] Figure 13 This is an internal structural diagram of another computer device provided in an embodiment of this application. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] The method for determining data tag objects provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located on a cloud or other network server. When a user logs into server 104 through terminal 102, the data tag object recommendation process is activated. Terminal 102 sends a business request to server 104, which can be a login request or other business processing request. Server 104 determines the data tag objects to recommend to terminal 102 based on the business request. Terminal 102 accesses the business content corresponding to the data tag object. The user likes, comments, or performs other actions on the data tag object on terminal 102. Terminal 102 transmits the received user behavior data to a processing server via the network for processing. This processing server can be server 104 or another server. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. Portable wearable devices can be smartwatches, smart bracelets, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0062] Figure 2This is a flowchart illustrating the method for determining a data tag object provided in this application embodiment. Taking the application of this method to a terminal device or server as an example, it includes the following steps:

[0063] S201, Obtain target user object information of the target user object.

[0064] The target user object is the user object of the data tag object to be recommended. The target user object can be determined based on the specific technical scenario; for example, when a business request is received, the user object carried in the business request is identified as the target user object. In some embodiments, the target user object can be the user identifier carried in the business request.

[0065] Target user information refers to information related to the target user, such as all or a combination of the user's name, department, and job title.

[0066] When retrieving target user object information, various methods can be used. For example, the target user object may carry target user object information in the aforementioned business request. Another example is that user object information is stored in a database; the server extracts the target user object (e.g., user identifier) ​​from the business request, and then searches the database for the corresponding user object information to obtain the target user object information.

[0067] S202, based on the target user object information, query and obtain several similar user objects that are similar to the target user object, and obtain several first data tag objects that have the highest correlation with the several similar user objects.

[0068] When querying and retrieving several similar user objects that are similar to the target user object, one or more different methods can be used to obtain them, such as querying based on the user object information table, or obtaining them based on the user object similarity matrix.

[0069] Data tag objects are objects used to make recommendations to users. Taking a technology scenario in the financial field as an example, data tag objects can be, for example, financial products, loan information, or other objects that can be recommended to users; there are no specific limitations here.

[0070] The degree of association between a user object and a data tag object is used to represent the degree of association between the user object and the data tag object. In some embodiments, this degree of association can be reflected based on the rating relationship between the user object and the data tag object.

[0071] The number of first data tag objects is not limited in this application embodiment. For example, it can be a predetermined number of data tag objects that have the highest correlation with several similar user objects, or it can be a predetermined number of data tag objects that have the highest correlation among several similar user objects.

[0072] S203, determine a number of second data tag objects that are most relevant to the target user object, and obtain a number of third data tag objects that are similar to the number of second data tag objects.

[0073] The number of third data tag objects that are similar to the number of second data tag objects can be represented by the similarity between the data tag objects. The specific number of second data tag objects and the number of third data tag objects are not limited in the embodiments of this application.

[0074] S204, obtain the target user object feature vector, query the several data tag content vectors with the highest similarity to the target user object feature vector, and obtain several fourth data tag objects corresponding to the several data tag content vectors. The target user object feature vector is determined based on the word vector of the data tag object corresponding to the target user object.

[0075] The target user object feature vector is a feature vector used to represent the data label object related to the target user object.

[0076] The word vector of a data tag object refers to the vector obtained after vectorizing the content of the data tag object, and it is related to the content of the data tag object.

[0077] The specific number of fourth data tag objects is not limited in this application embodiment.

[0078] S205, based on several first data tag objects, several third data tag objects, and several fourth data tag objects, determine the target data tag object to be recommended to the target user object.

[0079] When determining the target data tag objects to be recommended to the target user based on several first data tag objects, several third data tag objects, and several fourth data tag objects, one or more different methods can be used. For example, collaborative filtering can be performed on several first data tag objects, several third data tag objects, and several fourth data tag objects to obtain the target data tag objects to be recommended to the target user; or the rating results of several first data tag objects, several third data tag objects, and several fourth data tag objects can be sorted in descending order, and the top certain number of data tag objects with the highest rating results can be determined as the target data tag objects to be recommended to the target user.

[0080] The method for determining data tag objects provided in this embodiment involves: obtaining target user object information of a target user object; obtaining several first data tag objects similar to the target user object based on the target user object information; determining several second data tag objects with the highest correlation to the target user object; obtaining several third data tag objects similar to the several second data tag objects; obtaining the target user object feature vector of the target user object; obtaining several fourth data tag objects similar to the data tag content vector of the target user object based on the target user object feature vector; and determining the target data tag object to be recommended to the target user object based on the several first data tag objects, the several third data tag objects, and the several fourth data tag objects. Compared to traditional technologies that rely on the popularity and novelty of data tags to recommend data tags to users with low accuracy, this embodiment uses target user information, several second data tag objects with the highest relevance to the target user, and data tag objects determined by the target user's feature vector. By considering user similarity, the relevance between user and data tag objects, the relevance between data tag objects, and the similarity between the target user's feature vector determined based on word vectors and the data tag content vector of the data tag objects, this embodiment determines the data tag objects to be recommended from multiple dimensions. This enables more accurate recommendations for target users and solves the problem of inaccurate data tag object recommendations in traditional technologies.

[0081] See Figure 3 , Figure 3 This is a flowchart illustrating the process of determining a first data tag object in one embodiment. This embodiment provides an implementation method for determining the first data tag object. Based on the above embodiment, S202 includes the following:

[0082] S2021, obtain the target data label vector corresponding to the target user object, and query several similar data label vectors that are similar to the target data label vector in the user object similarity matrix.

[0083] In this embodiment, the target data tag vector corresponding to the target user object is a data tag vector obtained by vectorizing each data tag object with the highest correlation to the target user object.

[0084] In some embodiments, obtaining the target data tag vector corresponding to the target user object may include: obtaining the name of the data tag object that is highly associated with the target user object, vectorizing the data tag object name, and obtaining the target data tag vector.

[0085] The user object similarity matrix is ​​a matrix used to characterize the similarity between the data label vectors of each user object, reflecting the similarity between each user object.

[0086] In some embodiments, see Figure 4 The method for obtaining the user object similarity matrix includes the following steps S2121 to S2321.

[0087] Step S2121: Based on the rating relationship between each user object and each data tag object, obtain the data tag objects associated with each user object; wherein, the degree of association between each user object and each data tag object can be determined according to the rating relationship between each user object and each data tag object.

[0088] Among them, the rating relationship between user objects and each data label object is used to characterize the degree of evaluation of each data label object by user objects.

[0089] In some embodiments, the rating relationship between user objects and various data tag objects can be obtained in the following ways.

[0090] First, obtain the user's interaction data with each data tag object, which may include data such as browsing time, number of visits, visit depth, (un)likes / favorites, and comment content / byte count.

[0091] Then, the acquired data is assigned values, for example, 4 points for a like and -3 points for a dislike. The scores of users for each data tag object are summarized to obtain the user's rating value for each data tag object. This rating value is used to represent the rating relationship of users for each data tag object.

[0092] In some embodiments, the obtained score value can be directly used as the correlation degree. In other embodiments, the obtained score value can be further processed to obtain the correlation degree, such as by normalizing each score value or performing other processing. The embodiments of this application do not make specific limitations.

[0093] Step S2221: For each user object, vectorize the data tag object associated with each user object to obtain the data tag vector corresponding to each user object.

[0094] As mentioned above, a user object may be associated with multiple data tag objects. For example, a user object may be associated with n data tag objects A1...An, each with a corresponding name. When vectorizing the data tag objects associated with the user object, the names of the n data tag objects associated with the user object can be vectorized to obtain the data tag vector of the user object.

[0095] The vectorization method can be any method, and this application does not impose any specific limitations on the embodiments.

[0096] Step S2321: Obtain the user object similarity matrix based on the similarity of the data tag vectors corresponding to each user object.

[0097] Similarity characterizes the degree of similarity between data tag vectors. In some embodiments, the similarity between data tag vectors corresponding to each user object can be calculated using cosine similarity.

[0098] After obtaining the similarity of the data tag vectors corresponding to each user object, a user object similarity matrix can be constructed based on the calculated similarity. In the user object similarity matrix, the similarity value between two data tag vectors corresponding to each element, for example, the value of the element in the 2nd row and 3rd column, represents the similarity between the data tag vectors corresponding to the 2nd user object and the 3rd user object.

[0099] S2022, determine the user objects corresponding to a number of similar data label vectors, wherein the number of similar user objects includes the user objects corresponding to the number of similar data label vectors.

[0100] As mentioned above, one user object corresponds to one data tag vector. Therefore, after obtaining several similar data tag vectors, the user object corresponding to these data tag vectors can be obtained.

[0101] S2023, in the data tag object rating table, query and determine the first data tag objects that have the highest rating relationship with several similar user objects. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0102] The data label object rating table is used to represent the rating relationship between each user object and each data label object, and it reflects the rating results of each user object for each data label object.

[0103] In some embodiments, see Figure 5 The method for establishing a data label object rating table may include the following steps S2123 to S2323.

[0104] Step S2123: Obtain batch behavior data, which includes the behavior data of each user object on each data tag object within a predetermined time range.

[0105] Among them, behavioral data refers to the behavioral score data corresponding to each user's comments, likes, favorites or shares on each data tag object.

[0106] The process of obtaining batch behavioral data can be done by retrieving it from a database, which stores the behavioral data of each user object's comments, likes, favorites, or shares on each data tag object.

[0107] Step S2223: Summarize the batch behavior data using the user object as the primary key to obtain the summarized behavior data table.

[0108] Specifically, using the user object as the primary key, the batch behavior data of each data tag object is summarized to obtain the behavior data of the user object for each data tag object, resulting in a summarized behavior data table.

[0109] Step S2323: Perform principal component analysis on the summarized behavioral data table to obtain the data label object rating table, which includes the rating results of each user object for each data label object.

[0110] One way to perform principal component analysis on the summarized behavioral data table is to use the PCA principal component analysis algorithm to extract principal components from multiple behavioral data in the summarized behavioral data table.

[0111] In this embodiment, the data tag object rating table can be displayed in the form of a table including user object User_id, data tag object Item_id and rating result Score, or it can be displayed in the form of a chart.

[0112] In some embodiments, querying and retrieving several similar user objects that are similar to the target user object may further include:

[0113] Query the user object information table to obtain user object information that contains the department field information and the job position field information of the target user object information; the user object corresponding to the user object information is several similar user objects that are similar to the target user object.

[0114] The user object information table is used to represent information related to user objects. This table contains user attribute information related to user objects, such as user object ID, user object name, user object department, user object position, user asset information, and user image information.

[0115] The user object's department information includes the target user object's department field information, and the user object's job information includes the target user object's job field information.

[0116] Then, in the data tag object rating table, query and determine the first data tag objects that have the highest rating relationship with several similar user objects.

[0117] Therefore, in this embodiment, by including users in the same position and department as several similar user objects similar to the target user object, the number of similar user objects similar to the target user object can be further expanded, thereby improving the accuracy of finally determining the target data tag object to be recommended.

[0118] See Figure 6 , Figure 6 This is a flowchart illustrating the process of determining a third data tag object in one embodiment. This embodiment provides an implementation method for determining the third data tag object. Based on the above embodiment, S203 includes the following:

[0119] S2031, in the data tag object rating table, query and determine the several second data tag objects that have the highest rating relationship with the target user object.

[0120] The data tag object rating table can be the same as the data tag object rating table mentioned in the above embodiments. The number of second data tag objects is determined based on the rating results of the target user objects for each data tag object recorded in the data tag object rating table. It can be a predetermined number of data tag objects with the highest relevance to the target user objects, or it can be a predetermined proportion of data tag objects with the highest relevance to the target user objects.

[0121] S2032, obtain the user vectors corresponding to each of the several second data label objects, and query the data label object similarity matrix to obtain several similar user vectors that are similar to the user vectors.

[0122] In this embodiment, the user vector corresponding to each of the several second data tag objects is a user vector obtained by vectorizing the user object information of each user object that has the highest correlation with the several second data tag objects.

[0123] In some embodiments, obtaining the user vector corresponding to each of the several second data tag objects may include: obtaining the names of user objects that are highly associated with the several second data tag objects, vectorizing the user object names, and obtaining the user vectors.

[0124] The data tag object similarity matrix is ​​a matrix used to characterize the similarity between user vectors of each data tag object, reflecting the similarity between the data tag objects.

[0125] In some embodiments, see Figure 7 The method for obtaining the data label object similarity matrix includes the following steps S2132 to S2332.

[0126] Step S2132: Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object; wherein, the degree of association between each user object and each data tag object can be determined according to the rating relationship between each user object and each data tag object.

[0127] Among them, the rating relationship between user objects and each data label object is used to characterize the degree of evaluation of each data label object by user objects.

[0128] In some embodiments, the rating relationship between user objects and various data tag objects can be obtained in the following ways.

[0129] First, obtain the user's interaction data with each data tag object, which may include data such as browsing time, number of visits, visit depth, (un)likes / favorites, and comment content / byte count.

[0130] Then, the acquired data is assigned values, for example, 4 points for a like and -3 points for a dislike. The scores of users for each data tag object are summarized to obtain the user's rating value for each data tag object. This rating value is used to represent the rating relationship of users for each data tag object.

[0131] In some embodiments, the obtained score value can be directly used as the correlation degree. In other embodiments, the obtained score value can be further processed to obtain the correlation degree, such as by normalizing each score value or performing other processing. The embodiments of this application do not make specific limitations.

[0132] Step S2232: For each data tag object, vectorize the user object information of each user object of each data tag object to obtain the user vector corresponding to each data tag object.

[0133] As mentioned above, a data tag object may be associated with multiple user objects. For example, a data tag object may be associated with s user objects D1...Ds, each with a corresponding name. When vectorizing the user objects associated with the data tag object, the names of the s user objects associated with the data tag object may be vectorized to obtain the user vector of the data tag object.

[0134] The vectorization method can be any method, and this application does not impose any specific limitations on the embodiments.

[0135] Step S2332: Obtain the data label object similarity matrix based on the similarity of the user vectors corresponding to each data label object.

[0136] Similarity characterizes the degree of similarity between user vectors. In some embodiments, the similarity between user vectors corresponding to each data label object can be calculated using cosine similarity.

[0137] After obtaining the similarity of the user vectors corresponding to each data tag object, a data tag object similarity matrix can be constructed based on the calculated similarity. In the data tag object similarity matrix, each element represents the similarity value between two corresponding user vectors. For example, the value in the 3rd row and 4th column represents the similarity value between the 3rd and 4th data tag objects.

[0138] S2033, determine the data label objects corresponding to several similar user vectors, and obtain several third data label objects based on the data label objects corresponding to several similar user vectors.

[0139] As mentioned above, one data tag object corresponds to one user vector. Therefore, after obtaining several similar user vectors, the data tag objects corresponding to these user vectors can be obtained.

[0140] In this embodiment, the data tag objects corresponding to several similar user vectors can be determined by querying a data tag object rating table.

[0141] See Figure 8 , Figure 8 This is a flowchart illustrating the process of determining the fourth data tag object in one embodiment. This embodiment provides an implementation method for determining the fourth data tag object. Based on the above embodiment, S204 includes the following:

[0142] S2041, Obtain the target user object feature vector of the target user object.

[0143] The feature vector of the target user object is determined based on the word vector of the data tag object corresponding to the target user object.

[0144] Among them, the word vector of a data tag object refers to the vector used to represent the metadata information of the data tag object.

[0145] The word vectors of data tag objects can be determined in the following way: obtain the metadata information of the data tag object, perform vectorization processing on the metadata information, and obtain the word vectors of the data tag object.

[0146] S2042, in the data tag content similarity matrix, query and determine several data tag content vectors that are similar to the feature vector of the target user object.

[0147] The data tag content similarity matrix is ​​a matrix used to characterize the similarity between the data tag content vectors of each data tag object, reflecting the similarity between the data tag content vectors. The data tag content vectors can be word vectors of the data tag objects.

[0148] In some embodiments, see Figure 9 The method for obtaining the data tag content similarity matrix includes the following steps S2142 to S2342:

[0149] Step S2142: Obtain the metadata information of each data tag object.

[0150] Metadata information refers to the content information contained within a data tag object. In some embodiments, taking the application in the field of computer finance technology as an example, metadata information may include the business scope of the data tag object.

[0151] In some embodiments, metadata information can be obtained in the following ways: for example, metadata information can be obtained from a database that stores metadata information for each data tag object.

[0152] Step S2242: Vectorize the metadata information of each data tag object to obtain the word vector of each data tag object.

[0153] Taking a credit card as an example, the metadata information of each data tag object is vectorized for explanation. The metadata information of a credit card includes the available amount and the repayment date. The metadata information of each data tag object is vectorized to obtain a 100-dimensional vector [0.792, -0.177, ...]. This 100-dimensional vector [0.792, -0.177, ...] represents the word vector of the credit card.

[0154] The vectorization method can be any method, and this application does not impose any specific limitations on the embodiments.

[0155] Step S2342: Obtain the data tag content similarity matrix based on the similarity of the word vectors of each data tag object.

[0156] Similarity characterizes the degree of similarity between word vectors of various data label objects. In some embodiments, the similarity between word vectors of various data label objects can be calculated using cosine similarity.

[0157] S2043, obtain several fourth data tag objects corresponding to several data tag content vectors.

[0158] When obtaining several fourth data tag objects corresponding to several data tag content vectors, one or more different methods can be used. For example, the data tag content vectors can carry data tag object information. Another example is storing data tag objects in a database, calculating the similarity between the data tag content vectors and the data tag objects using cosine similarity, and then extracting the data tag object with the highest similarity to the data tag content vectors from the database as the fourth data tag object.

[0159] In this embodiment, the process of recommending data tag objects to target user objects is described.

[0160] refer to Figure 10 As shown, in the specific implementation process of this application embodiment, batch data loading, basic data determination, determination of recommended objects, and display of recommended objects are involved.

[0161] Before recommending data-tagged objects to target users, batch data loading and determination of basic data are performed first, and then the recommended objects are determined based on the determined basic data.

[0162] When loading batch data, batch data information can be retrieved from the database. This batch data can be data within a certain time period, specifically including: user object information, data tag objects, and user behavior data. Among them, the user object information of each user object includes the user object name, user object job information, and user object department information. The data tag object includes the data tag object name and the data tag object's metadata information. The user behavior data is the behavior data of the user object liking, commenting, collecting, or sharing the data tag object.

[0163] Then, for the batch data, its basic data is determined. This basic data includes: user object information table, data tag object rating table, user object similarity matrix, data tag object similarity matrix, and data tag content similarity matrix.

[0164] When determining the basic data, specifically: construct a user object information table based on the user object information. Each row in the table may include a user object name, the job information corresponding to the user object, and the department information corresponding to the user object.

[0165] Based on the user behavior data of each user object to each data tag object, the user behavior data is determined according to user behavior, such as liking, commenting, and collecting. Each user behavior corresponds to a different score, such as +4 for liking, +2 for commenting, and +5 for collecting. Principal component analysis is performed on the user behavior data of each user object to each data tag object to obtain principal user behavior data. Based on the principal user behavior data, the rating relationship of each user object to each data tag object is obtained, resulting in a data tag object rating table. In the table, each user object corresponds to at least one data tag object and the rating result of that user object for that data tag object.

[0166] In the data label object rating table, query the data label objects with high user object ratings, vectorize the data label objects with high user object ratings to obtain the data label vectors corresponding to the user objects, and use cosine similarity to calculate the similarity between the data label vectors corresponding to each user object to obtain the user object similarity matrix.

[0167] Query the user objects with high ratings in the data tag object rating table, vectorize the user objects with high ratings to obtain the user vectors corresponding to the data tag objects, and use cosine similarity to calculate the similarity between the user vectors corresponding to each data tag object to obtain the data tag object similarity matrix.

[0168] Obtain the metadata information of the data tag objects, perform vectorization processing on the metadata information of the data tag objects to obtain the word vectors of the data tag objects, and calculate the similarity between the word vectors of each data tag object to obtain the data tag content similarity matrix.

[0169] Once batch data is obtained, it can be stored for use in the subsequent process of determining the data label objects to be recommended.

[0170] The process of recommending data-labeled objects to target user objects is as follows: When a target user object logs into server 104 through terminal 102, the data-labeled object recommendation process is activated. Terminal 102 sends a business request to server 104, carrying the target user object information. Based on this target user object information, similar user objects similar to the target user object are queried from the user object information table or the user object similarity matrix. For example, users in the same position or department as the target user object can be obtained from the user object information table, or users with high similarity values ​​to the target user object can be obtained from the user object similarity matrix. In the data-labeled object scoring table, the data-labeled object with the highest score among the similar user objects is selected as the first data-labeled object.

[0171] In the data tag object rating table, the data tag object with the highest rating corresponding to the target user object is determined as the second data tag object; in the data tag object similarity matrix, the data tag object with a high similarity value to the second data tag object is determined as the third data tag object.

[0172] Obtain the word vector of the second data tag object as the target user object feature vector of the target user object; in the data tag content similarity matrix, determine the data tag content vector with a high similarity value to the target user object feature vector as the similar data tag content vector, and obtain the fourth data tag object based on the similar data tag content vector;

[0173] The similarity results of the first, third, and fourth data tag objects are sorted, and the object with the highest similarity is selected as the target data tag object, which is then recommended to the target user.

[0174] In this embodiment, the similarity between users in the same department and position is considered when recommending data tag objects. Users with similar preferences (i.e., high user similarity) are also recommended similar data tag objects. This fully considers users' needs and preferences for data tag objects, ensuring that both new and existing users receive suitable recommendations. User behavior data, i.e., a series of user actions related to data tag objects, is also considered. These actions reflect users' evaluations and preferences for the data tag objects, solving the problem of a single data source. Furthermore, several data tag objects are obtained through different methods, and collaborative filtering is used to finally determine the data tag objects to be recommended. This improves the comprehensiveness of user preference mining, effectively captures users' preferences for data tag objects, solves the cold start problem in data tag object recommendation methods, and provides users with more accurate data tag object recommendation services.

[0175] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0176] Based on the same inventive concept, this application also provides an apparatus for determining a data tag object to implement the method for determining a data tag object as described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more apparatus embodiments for determining a data tag object provided below can be found in the limitations of the method for determining a data tag object described above, and will not be repeated here.

[0177] See Figure 11 , Figure 11 This is a structural block diagram of a device for determining data tag objects provided in an embodiment of this application. The device 1100 includes: an information acquisition module 1101, a first object acquisition module 1102, a third object acquisition module 1103, a fourth object acquisition module 1104, and a recommended object determination module 1105, wherein:

[0178] Information acquisition module 1101 is used to acquire target user object information of the target user object;

[0179] The first object acquisition module 1102 is used to query and acquire several similar user objects that are similar to the target user object based on the target user object information, and acquire several first data tag objects that have the highest correlation with the several similar user objects.

[0180] The third object acquisition module 1103 is used to determine a number of second data tag objects that are most associated with the target user object, and to acquire a number of third data tag objects that are similar to the number of second data tag objects.

[0181] The fourth object acquisition module 1104 is used to acquire the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vector of the data tag object corresponding to the target user object.

[0182] The recommendation object determination module 1105 is used to determine the target data tag object to be recommended to the target user object based on a number of first data tag objects, a number of third data tag objects, and a number of fourth data tag objects.

[0183] The apparatus for determining data tag objects provided in this embodiment obtains target user object information of a target user object, and based on the target user object information, obtains several first data tag objects similar to the target user object; determines several second data tag objects with the highest relevance to the target user object, and obtains several third data tag objects similar to the second data tag objects; obtains the target user object feature vector of the target user object, and based on the target user object feature vector, obtains several fourth data tag objects similar to the data tag content vector of the target user object; and based on the several first data tag objects, several third data tag objects, and several fourth data tag objects, determines the target data tag object to be recommended to the target user object. Compared with the traditional technology that recommends data tag objects to users based on the popularity and novelty of data tag objects, which has low accuracy, the data tag objects determined in this embodiment through target user object information, several second data tag objects with the highest relevance to the target user object, and the target user object feature vector of the target user object can make more accurate recommendations for the target user object, solving the problem of inaccurate data tag object recommendations in the traditional technology.

[0184] Optionally, the first object acquisition module 1102 includes:

[0185] The data tag vector acquisition unit is used to acquire the target data tag vector corresponding to the target user object, and to query and obtain several similar data tag vectors that are similar to the target data tag vector in the user object similarity matrix.

[0186] The user object determination unit is used to determine the user object corresponding to a number of similar data label vectors, wherein the number of similar user objects includes the user object corresponding to the number of similar data label vectors.

[0187] Optional methods for obtaining the user object similarity matrix include:

[0188] Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object;

[0189] For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object;

[0190] Based on the similarity of the data tag vectors corresponding to each user object, a user object similarity matrix is ​​obtained.

[0191] Optionally, the first object acquisition module 1102 further includes:

[0192] The first object determination unit is used to query and determine the first data tag objects that have the highest rating relationship with several similar user objects in the data tag object rating table. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0193] Optional methods for creating the data label object rating table include:

[0194] Acquire batch behavior data, which includes: behavior data of each user object on each data tag object within a predetermined time range;

[0195] Summarize the batch behavior data using the user object as the primary key to obtain the summarized behavior data table;

[0196] Principal component analysis was performed on the summarized behavioral data table to obtain a data label object rating table, which includes the rating results of each user object for each data label object.

[0197] Optionally, the third object acquisition module 1103 includes:

[0198] The user vector acquisition unit is used to acquire user vectors corresponding to several second data label objects, and to query and obtain several similar user vectors that are similar to the user vectors in the data label object similarity matrix.

[0199] The third object determination unit is used to determine the data label objects corresponding to several similar user vectors, and to obtain several third data label objects based on the data label objects corresponding to several similar user vectors.

[0200] Optional methods for obtaining the data label object similarity matrix include:

[0201] Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object;

[0202] For each data tag object, the user object information of each user object of each data tag object is vectorized to obtain the user vector corresponding to each data tag object.

[0203] Based on the similarity of the user vectors corresponding to each data tag object, a data tag object similarity matrix is ​​obtained.

[0204] Optionally, the fourth object acquisition module 1104 includes:

[0205] The content vector determination unit is used to query and determine several data tag content vectors that are similar to the feature vectors of the target user object in the data tag content similarity matrix.

[0206] Optional methods for obtaining the data tag content similarity matrix include:

[0207] Retrieve metadata information for each data tag object;

[0208] Vectorize the metadata information of each data tag object to obtain the word vector of each data tag object;

[0209] Based on the similarity of word vectors of each data tag object, a data tag content similarity matrix is ​​obtained.

[0210] Each module in the aforementioned device for determining data tag objects can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0211] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 12As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores target user object information, data tag objects, and target user object feature vector data. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a method for determining data tag objects.

[0212] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 13 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a method for identifying data tag objects. The display screen can be an LCD screen or an e-ink display screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0213] Those skilled in the art will understand that Figure 12 and Figure 13 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0214] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method for determining a data tag object provided in the above embodiment.

[0215] Obtain target user object information;

[0216] Based on the target user object information, query and obtain several similar user objects that are similar to the target user object, and obtain several first data tag objects that have the highest correlation with the several similar user objects;

[0217] Identify several second data tag objects that are most relevant to the target user object, and obtain several third data tag objects that are similar to the several second data tag objects;

[0218] Obtain the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vectors of the data tag objects corresponding to the target user object.

[0219] Based on several first data tag objects, several third data tag objects, and several fourth data tag objects, determine the target data tag objects to be recommended to the target user objects.

[0220] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0221] Obtain the target data tag vector corresponding to the target user object, and in the user object similarity matrix, query to obtain several similar data tag vectors that are similar to the target data tag vector;

[0222] Identify the user objects corresponding to a number of similar data label vectors, where the number of similar user objects includes the user objects corresponding to the number of similar data label vectors.

[0223] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0224] Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object;

[0225] For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object;

[0226] Based on the similarity of the data tag vectors corresponding to each user object, a user object similarity matrix is ​​obtained.

[0227] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0228] In the data tag object rating table, the query identifies the first data tag objects that have the highest rating relationship with several similar user objects. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0229] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0230] Acquire batch behavior data, which includes: behavior data of each user object on each data tag object within a predetermined time range;

[0231] Summarize the batch behavior data using the user object as the primary key to obtain the summarized behavior data table;

[0232] Principal component analysis was performed on the summarized behavioral data table to obtain a data label object rating table, which includes the rating results of each user object for each data label object.

[0233] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0234] Obtain the user vector corresponding to each of the several second data label objects, and query the data label object similarity matrix to obtain several similar user vectors that are similar to the user vectors.

[0235] Identify data label objects corresponding to several similar user vectors, and based on these data label objects, obtain several third data label objects.

[0236] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0237] Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object;

[0238] For each data tag object, the user object information of each user object of each data tag object is vectorized to obtain the user vector corresponding to each data tag object.

[0239] Based on the similarity of the user vectors corresponding to each data tag object, a data tag object similarity matrix is ​​obtained.

[0240] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0241] In the data tag content similarity matrix, query and identify several data tag content vectors that are similar to the feature vector of the target user object.

[0242] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0243] Retrieve metadata information for each data tag object;

[0244] Vectorize the metadata information of each data tag object to obtain the word vector of each data tag object;

[0245] Based on the similarity of word vectors of each data tag object, a data tag content similarity matrix is ​​obtained.

[0246] The implementation principle and technical effects of the above embodiments are similar to those of the above method embodiments, and will not be repeated here.

[0247] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program being executed by a processor to implement the steps of the method for determining a data tag object provided in the above embodiment:

[0248] Obtain target user object information;

[0249] Based on the target user object information, query and obtain several similar user objects that are similar to the target user object, and obtain several first data tag objects that have the highest correlation with the several similar user objects;

[0250] Identify several second data tag objects that are most relevant to the target user object, and obtain several third data tag objects that are similar to the several second data tag objects;

[0251] Obtain the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vectors of the data tag objects corresponding to the target user object.

[0252] Based on several first data tag objects, several third data tag objects, and several fourth data tag objects, determine the target data tag objects to be recommended to the target user objects.

[0253] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0254] Obtain the target data tag vector corresponding to the target user object, and in the user object similarity matrix, query to obtain several similar data tag vectors that are similar to the target data tag vector;

[0255] Identify the user objects corresponding to a number of similar data label vectors, where the number of similar user objects includes the user objects corresponding to the number of similar data label vectors.

[0256] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0257] Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object;

[0258] For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object;

[0259] Based on the similarity of the data tag vectors corresponding to each user object, a user object similarity matrix is ​​obtained.

[0260] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0261] In the data tag object rating table, the query identifies the first data tag objects that have the highest rating relationship with several similar user objects. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0262] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0263] Acquire batch behavior data, which includes: behavior data of each user object on each data tag object within a predetermined time range;

[0264] Summarize the batch behavior data using the user object as the primary key to obtain the summarized behavior data table;

[0265] Principal component analysis was performed on the summarized behavioral data table to obtain a data label object rating table, which includes the rating results of each user object for each data label object.

[0266] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0267] Obtain the user vector corresponding to each of the several second data label objects, and query the data label object similarity matrix to obtain several similar user vectors that are similar to the user vectors.

[0268] Identify data label objects corresponding to several similar user vectors, and based on these data label objects, obtain several third data label objects.

[0269] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0270] Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object;

[0271] For each data tag object, the user object information of each user object of each data tag object is vectorized to obtain the user vector corresponding to each data tag object.

[0272] Based on the similarity of the user vectors corresponding to each data tag object, a data tag object similarity matrix is ​​obtained.

[0273] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0274] In the data tag content similarity matrix, query and identify several data tag content vectors that are similar to the feature vector of the target user object.

[0275] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0276] Retrieve metadata information for each data tag object;

[0277] Vectorize the metadata information of each data tag object to obtain the word vector of each data tag object;

[0278] Based on the similarity of word vectors of each data tag object, a data tag content similarity matrix is ​​obtained.

[0279] The implementation principle and technical effects of the above embodiments are similar to those of the above method embodiments, and will not be repeated here.

[0280] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method for determining a data tag object provided in the above embodiment:

[0281] Obtain target user object information;

[0282] Based on the target user object information, query and obtain several similar user objects that are similar to the target user object, and obtain several first data tag objects that have the highest correlation with the several similar user objects;

[0283] Identify several second data tag objects that are most relevant to the target user object, and obtain several third data tag objects that are similar to the several second data tag objects;

[0284] Obtain the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vectors of the data tag objects corresponding to the target user object.

[0285] Based on several first data tag objects, several third data tag objects, and several fourth data tag objects, determine the target data tag objects to be recommended to the target user objects.

[0286] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0287] Obtain the target data tag vector corresponding to the target user object, and in the user object similarity matrix, query to obtain several similar data tag vectors that are similar to the target data tag vector;

[0288] Identify the user objects corresponding to a number of similar data label vectors, where the number of similar user objects includes the user objects corresponding to the number of similar data label vectors.

[0289] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0290] Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object;

[0291] For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object;

[0292] Based on the similarity of the data tag vectors corresponding to each user object, a user object similarity matrix is ​​obtained.

[0293] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0294] In the data tag object rating table, the query identifies the first data tag objects that have the highest rating relationship with several similar user objects. The data tag object rating table includes the rating relationship of each user object to each data tag object.

[0295] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0296] Acquire batch behavior data, which includes: behavior data of each user object on each data tag object within a predetermined time range;

[0297] Summarize the batch behavior data using the user object as the primary key to obtain the summarized behavior data table;

[0298] Principal component analysis was performed on the summarized behavioral data table to obtain a data label object rating table, which includes the rating results of each user object for each data label object.

[0299] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0300] Obtain the user vector corresponding to each of the several second data label objects, and query the data label object similarity matrix to obtain several similar user vectors that are similar to the user vectors.

[0301] Identify data label objects corresponding to several similar user vectors, and based on these data label objects, obtain several third data label objects.

[0302] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0303] Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object;

[0304] For each data tag object, the user object information of each user object of each data tag object is vectorized to obtain the user vector corresponding to each data tag object.

[0305] Based on the similarity of the user vectors corresponding to each data tag object, a data tag object similarity matrix is ​​obtained.

[0306] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0307] In the data tag content similarity matrix, query and identify several data tag content vectors that are similar to the feature vector of the target user object.

[0308] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0309] Retrieve metadata information for each data tag object;

[0310] Vectorize the metadata information of each data tag object to obtain the word vector of each data tag object;

[0311] Based on the similarity of word vectors of each data tag object, a data tag content similarity matrix is ​​obtained.

[0312] The implementation principles and technical effects of the above embodiments are similar to those of the above method embodiments, and will not be repeated here. It should be noted that the data involved in this application (including but not limited to data used for analysis, stored data, and displayed data) are all information and data authorized by the user or fully authorized by all parties.

[0313] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0314] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0315] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining data tag objects, characterized in that, The method includes: Obtain target user object information; Based on the target user object information, obtain the target data tag vector corresponding to the target user object; in the user object similarity matrix, query and obtain several similar data tag vectors that are similar to the target data tag vector; determine the user object corresponding to the several similar data tag vectors, the several similar user objects include the user objects corresponding to the several similar data tag vectors; in the data tag object rating table, query and determine several first data tag objects with the highest rating relationship with the several similar user objects, the data tag object rating table includes the rating relationship of each user object to each data tag object; Determine a number of second data tag objects that have the highest correlation with the target user object, obtain the user vector corresponding to each of the number of second data tag objects, query the data tag object similarity matrix to obtain a number of similar user vectors that are similar to the user vectors, determine the data tag objects corresponding to the number of similar user vectors, and obtain a number of third data tag objects based on the data tag objects corresponding to the number of similar user vectors. Obtain the target user object feature vector, query the data tag content vectors with the highest similarity to the target user object feature vector, and obtain the fourth data tag objects corresponding to the data tag content vectors. The target user object feature vector is determined based on the word vectors of the data tag objects corresponding to the target user object. Based on the plurality of first data tag objects, the plurality of third data tag objects, and the plurality of fourth data tag objects, a target data tag object to be recommended to the target user object is determined; The method for establishing the data label object rating table includes: Acquire batch behavioral data, which includes: behavioral data of each user object on each data tag object within a predetermined time range; summarize the batch behavioral data with user objects as the primary key to obtain a summarized behavioral data table; perform principal component analysis on the summarized behavioral data table to obtain the data tag object rating table, which includes the rating results of each user object on each data tag object; the behavioral data refers to the behavioral score data corresponding to each user object's comment, like, favorite, or share behavior on each data tag object.

2. The method according to claim 1, characterized in that, The methods for obtaining the user object similarity matrix include: Based on the rating relationship between each user object and each data tag object, obtain the data tag object associated with each user object; For each user object, the data tag object associated with each user object is vectorized to obtain the data tag vector corresponding to each user object; The user object similarity matrix is ​​obtained based on the similarity of the data tag vectors corresponding to each user object.

3. The method according to claim 1, characterized in that, The methods for obtaining the data tag object similarity matrix include: Based on the rating relationship between each user object and each data tag object, obtain each user object associated with each data tag object; For each of the data tag objects, the user object information of each user object of each of the data tag objects is vectorized to obtain the user vector corresponding to each of the data tag objects; The similarity matrix of the data tag objects is obtained based on the similarity of the user vectors corresponding to each data tag object.

4. The method according to any one of claims 1-2, characterized in that, The query selects the data tag content vectors that have the highest similarity to the feature vector of the target user object, including: In the data tag content similarity matrix, several data tag content vectors that are similar to the feature vector of the target user object are queried and identified.

5. The method according to claim 4, characterized in that, The methods for obtaining the data tag content similarity matrix include: Retrieve metadata information for each data tag object; The metadata information of each data tag object is vectorized to obtain the word vector of each data tag object; The content similarity matrix of the data tags is obtained based on the similarity of the word vectors of each data tag object.

6. An apparatus for determining a data tag object, characterized in that, The device includes: The information acquisition module is used to acquire target user object information. The first object acquisition module is configured to: acquire a target data tag vector corresponding to the target user object based on the target user object information; query a number of similar data tag vectors similar to the target data tag vector in the user object similarity matrix; determine the user object corresponding to the number of similar data tag vectors, wherein the number of similar user objects includes the user object corresponding to the number of similar data tag vectors; and query a number of first data tag objects with the highest rating relationship to the number of similar user objects in a data tag object rating table, wherein the data tag object rating table includes the rating relationship of each user object to each data tag object. The third object acquisition module is used to determine a number of second data tag objects with the highest correlation to the target user object, obtain the user vector corresponding to each of the number of second data tag objects, query and obtain a number of similar user vectors similar to the user vectors in the data tag object similarity matrix, determine the data tag objects corresponding to the number of similar user vectors, and obtain a number of third data tag objects based on the data tag objects corresponding to the number of similar user vectors. The fourth object acquisition module is used to acquire the target user object feature vector of the target user object, query several data tag content vectors with the highest similarity to the target user object feature vector, and obtain several fourth data tag objects corresponding to the several data tag content vectors. The target user object feature vector is determined based on the word vector of the data tag object corresponding to the target user object. The recommendation object determination module is used to determine the target data tag object to be recommended to the target user object based on the plurality of first data tag objects, the plurality of third data tag objects, and the plurality of fourth data tag objects; The method for establishing the data label object rating table includes: Acquire batch behavioral data, which includes: behavioral data of each user object on each data tag object within a predetermined time range; summarize the batch behavioral data with user objects as the primary key to obtain a summarized behavioral data table; perform principal component analysis on the summarized behavioral data table to obtain the data tag object rating table, which includes the rating results of each user object on each data tag object; the behavioral data refers to the behavioral score data corresponding to each user object's comment, like, favorite, or share behavior on each data tag object.

Images