User Identification Method, Apparatus, Device, and Storage Medium

Abstract

An embodiment of the present application discloses a method, device, equipment, and storage medium for user identification. The method includes: obtaining a sample user set for a target service, target application programs installed by sample users in the sample user set, object attributes of sample users, and a reference application program; counting a first installation ratio corresponding to the reference application program installed by sample users in the sample user set, and performing an association category division on the association relationship between the reference application program and the target service according to the first installation ratio to obtain a reference association category; if the target application program matches the reference application program, determining the reference association category as the target association category corresponding to the target application program; and adjusting a user identification model by using the target association category and the object attributes of sample users to obtain a target user identification model. Through the present application, the discrimination degree of application programs can be improved, and the identification accuracy of the user identification model can be improved.

Description

Technical Field

[0001] This application relates to the technical field of artificial intelligence - machine learning, and particularly to a user identification method, apparatus, device, and storage medium. Background Art

[0002] With the rapid development of Internet technology, more and more users participate in online activities, resulting in a massive increase in user behavior data on the Internet, and this behavior data has become increasingly valuable. For example, the behavior data of the application programs installed by users is used to train a user identification model, and based on the identification results obtained from the user identification model, content that the user is interested in, such as online courses, games, audio - video memberships, and coupons, etc., is recommended to the user.

[0003] Practice has found that in the process of training a user identification model, all business scenarios train the user identification model with general features, resulting in low identification accuracy of the user identification model. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of this application is to provide a user identification method, apparatus, device, and storage medium, which can improve the discrimination of application programs and the identification accuracy of the user identification model.

[0005] On the one hand, an embodiment of this application provides a user identification method, including:

[0006] Obtain a sample user set for a target business, the target application programs installed by the sample users in the sample user set, the object attributes of the sample users, and reference application programs having an association relationship with the target business;

[0007] Statistically calculate a first installation ratio corresponding to the installation of the reference application program by the sample users in the sample user set, and perform an association category division on the association relationship between the reference application program and the target business according to the first installation ratio to obtain a reference association category;

[0008] If the target application program matches the reference application program, determine that there is an association relationship between the target application program and the target business, and use the reference association category as the target association category corresponding to the association relationship between the target application program and the target business;

[0009] Adjust the user identification model by using the target association category and the object attributes of the sample users to obtain a target user identification model for identifying target users associated with the target business.

[0010] On the one hand, an embodiment of this application provides a user identification apparatus, including:

[0011] An acquisition module, configured to acquire a set of sample users for a target service, a target application installed by the sample users in the set of sample users, object attributes of the sample users, and a reference application having an association with the target service;

[0012] A division module, configured to count a first installation ratio corresponding to the reference application installed by the sample users in the set of sample users, and perform an association category division on the association between the reference application and the target service according to the first installation ratio, so as to obtain a reference association category;

[0013] A determination module, configured to determine that there is an association between the target application and the target service if the target application matches the reference application, and use the reference association category as the target association category corresponding to the association between the target application and the target service;

[0014] An adjustment module, configured to adjust a user recognition model by using the target association category and the object attributes of the sample users, so as to obtain a target user recognition model for recognizing target users associated with the target service.

[0015] On the one hand, the present application provides a computer device, including: a processor and a memory;

[0016] Wherein, the above-mentioned memory is used to store a computer program, and the above-mentioned processor is used to call the above-mentioned computer program to execute the following steps:

[0017] Acquire a set of sample users for a target service, a target application installed by the sample users in the set of sample users, object attributes of the sample users, and a reference application having an association with the target service;

[0018] Count a first installation ratio corresponding to the reference application installed by the sample users in the set of sample users, and perform an association category division on the association between the reference application and the target service according to the first installation ratio, so as to obtain a reference association category;

[0019] If the target application matches the reference application, determine that there is an association between the target application and the target service, and use the reference association category as the target association category corresponding to the association between the target application and the target service;

[0020] Adjust a user recognition model by using the target association category and the object attributes of the sample users, so as to obtain a target user recognition model for recognizing target users associated with the target service.

[0021] An embodiment of the present application provides a computer-readable storage medium on the one hand. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, the following steps are performed:

[0022] Obtain a set of sample users for a target service, a target application installed by the sample users in the set of sample users, object attributes of the sample users, and a reference application having an association with the target service;

[0023] Statistically calculate a first installation ratio corresponding to the installation of the reference application by the sample users in the set of sample users, and classify the association between the reference application and the target service according to the first installation ratio to obtain a reference association category;

[0024] If the target application matches the reference application, determine that there is an association between the target application and the target service, and use the reference association category as the target association category corresponding to the association between the target application and the target service;

[0025] Adjust a user recognition model by using the target association category and the object attributes of the sample users to obtain a target user recognition model for identifying target users associated with the target service.

[0026] In the present application, first, an electronic device can statistically calculate a first installation ratio corresponding to the installation of a reference application by sample users, and classify the association between the application on the market (i.e., the reference application) and the target service according to the first installation ratio to obtain a reference association category. Then, when the target application matches the reference application, it is determined that there is an association between the target application and the target service, and the reference association category is used as the target association category corresponding to the association between the target application and the target service. By matching the target application with the reference application, the target association category corresponding to the target application can be obtained without calculating the corresponding first installation ratio for the target application installed by each sample user, which can reduce the amount of computation and improve the efficiency of obtaining the target association category corresponding to the target application. Further, the target association category and the object attributes of the sample users can be used as features to train the user recognition model to obtain a target user recognition model; in different application scenarios, the reference association category corresponding to the reference application is different, so that the target association category corresponding to the target application is also different. That is to say, the present application can dynamically and adaptively construct the features of the user recognition model according to the business scenario, which can improve the discrimination between applications and thus improve the recognition accuracy of the user recognition model. Description of the Drawings

[0027] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0028] Figure 1 is a schematic structural diagram of a user identification system provided by the present application;

[0029] Figure 2a is a schematic diagram of an application program category tree provided by the present application;

[0030] Figure 2b is a schematic diagram of an application program category tree provided by the present application;

[0031] Figure 3a is a schematic diagram of a data interaction scenario provided by the present application;

[0032] Figure 3b is a schematic diagram of a data interaction scenario provided by the present application;

[0033] Figure 4 is a schematic flowchart of a user identification method provided by the present application;

[0034] Figure 5 is a schematic diagram of a scenario for training a user identification model provided by the present application;

[0035] Figure 6 is a schematic diagram of a scenario for determining a target associated category through an application program category tree provided by the present application;

[0036] Figure 7 is a schematic diagram of a scenario for determining a target associated category through an application program category tree provided by the present application;

[0037] Figure 8 is a schematic diagram of a scenario for training a user identification model provided by the present application;

[0038] Figure 9 is a schematic structural diagram of a user identification device provided by an embodiment of the present application;

[0039] Figure 10 is a schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0040] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present application.

[0041] Artificial Intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, a theory, method, technology, and application system that can perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling the machines to have the functions of perception, reasoning, and decision-making.

[0042] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level technologies and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, large user identification technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0043] Among them, Machine Learning (ML) is an interdisciplinary subject in multiple fields, involving multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how a computer simulates or implements human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve its own performance. Machine learning is the core of artificial intelligence and the fundamental way to make a computer intelligent. Its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and rote learning.

[0044] The user identification method provided by the embodiments of this application mainly relates to artificial intelligence - machine learning technology, that is, by analyzing the target application installed by the sample user, to obtain the target association category corresponding to the association relationship between the target application and the target business, and using the target association category and the object attributes of the sample user as features to train the user identification model, and obtaining the target user identification model for identifying the target user associated with the target business. It can be seen that the target association category is dynamically and adaptively constructed according to the business scenario, which can improve the discrimination of the application, and further improve the identification accuracy of the user identification model.

[0045] To more clearly explain this application, first introduce the user identification system used to implement the user identification method of this application, as Figure 1 shown, the user identification system includes a server 10 and multiple terminals. For example, Figure 1 taking four terminals as an example, namely terminal 11, terminal 12, terminal 13, and terminal 14. Each terminal is respectively connected to the server 10 through a network, so that each terminal can communicate with the server 10 through the network connection.

[0046] Among them, the server 10 can refer to the backend device used to provide applications for users. In this application, the applications provided by the server can be called candidate applications, that is, candidate applications can refer to the applications available on the market, and candidate applications can be downloaded and installed by users; candidate applications can include game applications, payment applications, shopping applications, multimedia applications (such as audio and video applications), and educational applications, etc. The server 10 can also be used to record the target applications downloaded and installed by the sample user from the server 10, that is, the target application can refer to any application in the applications installed by the sample user; further, the server 10 can be used to analyze the target applications installed by the sample user, and use the analysis results and the object attributes of the sample user as features to train the user identification model, and obtain the target user identification model for identifying the target user associated with the target business. The server 10 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0047] It is understandable that the candidate application programs in the server 10 that have an associated relationship with the target service can be called reference application programs. That is, the reference application programs can refer to the application programs that have significant features relative to the target service. In other words, the reference application programs refer to the application programs that have a positive or negative impact on the target service. Here, the target service can refer to the service scenario to which this application is applied. That is, the target service can include game service scenarios, education service scenarios, credit overdue service scenarios, and so on.

[0048] It is understandable that, for the convenience of distinction, the association category corresponding to the association relationship between the reference application program and the target service is called the reference association category. This reference association category includes positive correlation and negative correlation. Positive correlation means that the reference application program has a positive impact on the target service. That is, the probability that a user who installs a reference application program with a positive correlation with the target service becomes a target user of the target service is relatively low. Negative correlation means that the reference application program has a negative impact on the target service. That is, the probability that a user who installs a reference application program with a negative correlation with the target service becomes a target user of the target service is relatively low.

[0049] Optionally, the server 10 can determine the service relationship between the reference application program and the target service according to the service keywords of the target service. The service relationship includes the same category relationship and the non - same category relationship. The same category relationship means that the reference application program and the target service belong to the same service scenario. Non - same category means that the reference application program and the target service do not belong to the same service scenario. For example, in the education service scenario, the service keywords can include training, learning, teaching, etc. If the reference application program includes service keywords such as learning or training, it is said that the reference application program belongs to the education service scenario. If the reference application program does not include service keywords such as learning or training, it is said that the reference application program does not belong to the education service scenario. The service relationship between the reference application program and the target service can be used to explain the recognition result output by the user recognition model.

[0050] The application program category tree in this application is a network used to describe the association category corresponding to the association relationship between the reference application program and the target service, as well as the service relationship between the reference application program and the target service. This application program category tree includes a root node and multiple leaf nodes. The root node refers to the node used to store the reference application programs that have an associated relationship with the target service. That is, the root node is used to store the application programs with significance. The leaf nodes are used for reference application programs with different association categories.

[0051] For example, as Figure 2aAs shown in the figure, the application category tree 1 includes three layers. The first layer includes a root node 15 for storing all reference applications. The second layer includes a first leaf node 16 and a second leaf node 17. The first leaf node 16 is used to store reference applications whose reference association category corresponding to the association relationship with the target service is positively correlated. The second leaf node 17 is used to store reference applications whose reference association category corresponding to the association relationship with the target service is negatively correlated. The third layer includes a third leaf node 18 and a fourth leaf node 19. The third leaf node 18 is used to store reference applications whose reference association category corresponding to the association relationship with the target service in the second leaf node is positively correlated at the first level. The fourth leaf node 19 is used to store reference applications whose reference association category corresponding to the association relationship with the target service in the second leaf node is positively correlated at the second level.

[0052] Optionally, as Figure 2b shown in the figure, the application category tree 2 includes four layers. The first layer includes a root node 20 for storing all reference applications. The second layer includes a first leaf node 21 and a second leaf node 22. The first leaf node 21 is used to store reference applications whose reference association category corresponding to the association relationship with the target service is positively correlated. The second leaf node 22 is used to store reference applications whose reference association category corresponding to the association relationship with the target service is negatively correlated. The third layer includes a third leaf node 23, a fourth leaf node 24, a fifth leaf node 25, and a sixth leaf node 26. The third leaf node 23 is used to store reference applications whose business relationship with the target service in the first leaf node 21 is of the same category. The fourth leaf node 24 is used to store reference applications whose business relationship with the target service in the first leaf node 21 is of the same category. The fifth leaf node 25 is used to store reference applications whose business relationship with the target service in the second leaf node 22 is of the same category. The sixth leaf node 26 is used to store reference applications whose business relationship with the target service in the second leaf node 22 is of the same category. The fourth layer includes a seventh leaf node 27 and an eighth leaf node 28. The seventh leaf node 27 is used to store reference applications whose reference association category corresponding to the association relationship with the target service in the third leaf node is positively correlated at the first level. The eighth leaf node 28 is used to store reference applications whose reference association category corresponding to the association relationship with the target service in the third leaf node is positively correlated at the second level. Of course, the reference applications in the fourth leaf node, the fifth leaf node, and the sixth leaf node can be divided with reference to the third leaf node to obtain more leaf nodes, which will not be elaborated here.

[0053] Among them, the terminal can refer to a user-facing device. The terminal can be a device used by a sample user to download and install a target application from a server. That is, the target application here can refer to any one of the above candidate applications. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a smart vehicle connection, a smart TV, etc., but is not limited thereto. Each terminal and the server can be directly or indirectly connected through wired or wireless communication methods, and this application does not make restrictions here.

[0054] The object attributes in this application can include age, gender, household register, etc. The above sample user can refer to the user to whom the data for training the user recognition model belongs. The sample user can include positive sample users and negative sample users. The positive sample user can refer to the target population, and the negative sample user can refer to the non-target population. The target population can refer to the set of users concerned in the business scenario. In this application, the positive sample user can refer to a user interested in a certain business, and the non-sample user refers to a user not interested in a certain business. For example, in the game business scenario, the positive sample user can be a user interested in the game, such as a user who installs the game application; the negative sample user refers to a user not interested in the game, such as a user who does not install the game application.

[0055] For ease of understanding, further, please refer to Figure 3a and Figure 3b , which is a schematic diagram of a data interaction scenario provided by an embodiment of this application. Among them, as Figure 3a and Figure 3b shown, the application server can be the above server 10, and as Figure 3a and Figure 3b shown, the terminal can be any one of the terminals 11 to 14 in the corresponding embodiment of the above Figure 1 . For example, the terminal can be the above terminal 11.

[0056] As Figure 3a and Figure 3b shown, taking the education business scenario as an example for illustration, when it is necessary to push a certain programming online course to a user, in order to achieve accurate push and reduce the push cost, it is necessary to obtain the target users interested in the programming online course and only recommend the programming network program to the target users, which can improve the push effect and reduce the push cost.

[0057] Specifically, in S1, the server obtains a set of sample users for the programming online course. The sample users in this set of sample users can refer to users who have downloaded and installed the application on the server. The set of sample users includes positive sample users and negative sample users. Positive sample users can refer to users who have purchased the programming online course or installed programming applications. Correspondingly, negative sample users can refer to users other than positive sample users in the set of sample users.

[0058] In S2, the server builds an application category tree.

[0059] a. The server can obtain the installation shares of candidate applications among positive sample users and negative sample users. The installation share of a candidate application among positive sample users refers to the proportion of positive sample users who have installed the candidate application. The installation share of a candidate application among positive sample users can be calculated according to the following formula (1).

[0060]

[0061] Among them, in formula (1), P1 represents the installation share of the candidate application installed by positive sample users, Z represents the number of positive sample users who have installed the candidate application, that is, the installation volume of the candidate application among positive sample users, and R1 refers to the number of positive sample users. Similarly, the installation share of a candidate application among negative sample users refers to the proportion of negative sample users who have installed the candidate application. The installation share of a candidate application among negative sample users can be calculated according to the following formula (2).

[0062]

[0063] Among them, in formula (2), P2 represents the installation share of the candidate application installed by negative sample users, H represents the number of negative sample users who have installed the candidate application, that is, the installation volume of the candidate application among negative sample users, and R2 refers to the number of negative sample users.

[0064] b. The server can determine the installation proportion corresponding to the installation of the candidate application by sample users based on the installation shares of the candidate application among positive sample users and negative sample users. This installation proportion can be called the significance index, and this significance index can be calculated according to the following formula (3).

[0065]

[0066] Among them, in formula (2), Q represents the installation proportion corresponding to the installation of the candidate application by sample users.

[0067] c. Select candidate applications with significant features as reference applications and add the reference applications to the root node of the application category tree. The larger the value of Q, the more users among the positive sample users install the candidate application; the smaller the value of Q, the fewer users among the positive sample users install the candidate application. That is, when the significance index is greater than the first installation ratio threshold, it indicates that the candidate application has a positive impact on the target business; when the significance index is less than the second installation ratio threshold, it indicates that the candidate application has a negative impact on the target business; when the significance index is less than or equal to the first installation ratio threshold and greater than or equal to the second installation ratio threshold, it indicates that the candidate application has a relatively small impact on the target business or no impact and can be ignored. Therefore, candidate applications with a significance index greater than the first installation ratio threshold and candidate applications with a significance index less than the second installation ratio threshold can be selected as reference applications with significant features and added to the root node of the application category tree. Among them, the first installation ratio threshold is greater than the second installation ratio threshold. Optionally, the first installation ratio threshold can be greater than 100, such as 120, and the second installation ratio threshold can be the ratio between 100 and the first installation ratio threshold.

[0068] d. Configure the second layer of the application category tree, which includes the first leaf node and the second leaf node. Add the reference applications with a significance index greater than the first installation ratio threshold to the first leaf node, and determine that the association category corresponding to the association relationship between the reference applications in the first leaf node and the target business is positively correlated; add the reference applications with a significance index less than the second installation ratio threshold to the second leaf node, and determine that the association category corresponding to the association relationship between the reference applications in the second leaf node and the target business is negatively correlated.

[0069] e. Based on the second layer of the application category tree, split the reference applications into those with the same category relationship with the target business and those with a non - same category relationship with the target business according to whether the reference applications contain business keywords related to the target business, and configure the third layer of the application category tree based on this. The business keywords can be customized. In the education business scenario, the business keywords can include keywords such as "learning" and "training". The configuration of the second layer can be referred to Figure 2b , which will not be elaborated here.

[0070] d. Configure the fourth layer of the application category tree, and sort the reference applications in the third leaf node according to the significance index corresponding to the reference application. For example, sort the reference applications in the third leaf node in descending order according to the significance index corresponding to the reference application. According to the sorting order, cumulatively calculate the coverage rate of each reference application in turn to obtain the sum of the coverage rates. Divide the reference applications in the third leaf node into multiple groups according to the sum of the coverage rates, and the sum of the coverage rates corresponding to the reference applications in each group is greater than the coverage rate threshold. For example Figure 2b In it, the reference applications in the third leaf node will be divided into two groups according to the sum of the coverage rates. This grouping includes the first group and the second group. Add the reference applications belonging to the first group to the seventh leaf node, and add the reference applications belonging to the second group to the eighth leaf node. Among them, the sum of the coverage rates corresponding to the reference applications within each group is greater than the first coverage rate threshold and less than the second coverage rate threshold. The reference association category corresponding to the association relationship between the reference applications in the seventh leaf node and the target service is positively correlated at the first level, and the reference association category corresponding to the association relationship between the reference applications in the eighth leaf node and the target service is positively correlated at the second level. Among them, the coverage rate of the reference application can refer to the ratio of the installation volume corresponding to the installation of the reference application by the sample users to the total number of sample users.

[0071] e. After obtaining the application category tree, the number of applications belonging to each leaf node of the application category tree in the target application downloaded by the sample users can be queried.

[0072] s3. Train the user recognition model with the number of applications and the object attributes of the sample users to obtain the target user recognition model.

[0073] s4. The server spreads the target users. The server can query the number of applications belonging to the leaf nodes of the application category tree in the historical applications installed by the candidate users in the past, and the association category corresponding to the association relationship between the historical applications and the target service. Use the target user recognition model to identify the number of historical applications and the association category corresponding to the association relationship between the historical applications and the target service to obtain the probability of each candidate user. This probability is used to reflect the degree of association between the candidate user and the target service; that is, the greater the probability, the greater the degree of association; the smaller the probability, the smaller the degree of association. Select the target users corresponding to the target service from the candidate users according to the requirements and the probability; among them, the candidate users can refer to all users in the application list, and the above sample users can refer to some users in the application list. The application list includes users who have downloaded the applications in the server.

[0074] For example, such as Figure 3bIn this case, the user can configure the target service on the user interface 29 of the server, such as the number of users to be spread for the target service (i.e., select the head magnitude), the output user ID type (phone number, login account, etc.), the service name (i.e., the task name), the ID of the user recognition model, and so on. After creating the target service, the server can select the target users belonging to the target service according to the historical application corresponding association categories of the target users and the object attributes.

[0075] Optionally, after obtaining the target users for spreading, the server can output Table 1 on the user interface. Table 1 includes the service name, recognition status, ID of the user recognition model, delivery magnitude, and operation options related to the target service of the target users of the target service. The operation options related to the target service include options such as download, delete, copy, view details, resubmit, etc. The download option is used to download the list of target users belonging to the target service. The delete option is used to delete the information related to the target service (such as the list of target users); the copy option is used to copy the information related to the target service; the resubmit option is used to trigger the user recognition model to re-obtain the target users corresponding to the target service; the delivery magnitude refers to the number of users to whom the content related to the target service is pushed each time. Table 1 can also include information such as the customer name, customer type, service creator, take the head magnitude, and creation time to which the target service belongs; taking the head magnitude refers to the total number of users corresponding to the target users associated with the target service. As can be seen from Table 1, the model ID used in business scenario 1 is 241, and the model ID used in business scenario 2 is 242, that is, different business scenarios use different user recognition models, which can improve the accuracy of obtaining the target users corresponding to the target service.

[0076] Table 1:

[0077]

[0078]

[0079] Optionally, after the server obtains the target users, it can send the user list including the target users to the terminal, and the terminal pushes the content related to the target service to the target users; or, the server can directly push the content related to the target service to the target users. Such as Figure 3a and Figure 3b In, the server can push programming network courses to the target users. It should be noted that the above steps s1~s4 can be implemented by the server, or can be implemented by the terminal; or can be implemented by the cooperation of the terminal and the server. For example, the server can implement steps s1~s3, and the terminal can implement step s4.

[0080] Furthermore, please refer to Figure 4, is a schematic flow chart of a user identification method provided by an embodiment of the present application. As Figure 4 shown, this method can be executed by an electronic device, that is, the electronic device can be the Figure 1 server in Figure 1 , that is, this method can be executed by the server; or, the electronic device can be the Figure 1 terminal in

[0081] S101, Obtain a sample user set for the target service, the target application installed by the sample users in the sample user set, the object attributes of the sample users, and the reference application that has an association relationship with the target service.

[0082] Specifically, the electronic device can obtain an application list from an application supply device (such as the above-mentioned server). The application list includes users who have downloaded the applications in the application supply device, and information such as the target application downloaded and installed by each user. Some users can be selected from the application list as sample users. The sample users include positive sample users and negative sample users. The positive sample users refer to the sample users marked as associated with the target service, and the negative sample users refer to the sample users marked as not associated with the target service. And screen out the reference applications that have an association relationship with the target service from the applications provided by the application supply device, that is, the reference applications refer to the applications with significant features.

[0083] S102, Statistically calculate the first installation ratio corresponding to the installation of the reference application by the sample users in the sample user set.

[0084] Specifically, the electronic device can, according to the applications installed by the sample users, statistically calculate the installation ratio corresponding to the installation of the reference application by the positive sample users in the sample user set and the installation ratio corresponding to the installation of the reference application by the negative sample users in the sample user set, and determine the first installation ratio according to the installation ratio corresponding to the positive sample users in the sample set and the installation ratio corresponding to the negative sample users.

[0085] S103, Perform an association category division on the association relationship between the reference application and the target service according to the first installation ratio to obtain a reference association category.

[0086] Specifically, the larger the first installation proportion is, it indicates that the installation proportion of the reference application among the positive sample users is relatively high, that is, the positive sample users prefer to install the reference application, that is, the reference application has a positive impact on the target business; if the first installation proportion is smaller, it indicates that the installation proportion of the reference application among the negative sample users is relatively high, that is, the negative sample users prefer to install the reference application, that is, the reference application has a negative impact on the target business. Therefore, the association relationship between the reference application and the target business can be classified according to the first installation proportion to obtain the reference association category; that is, the association relationship between the reference application and the target business is segmented according to the first installation proportion to obtain the reference association category. That is to say, for different business scenarios, the reference association category corresponding to the reference application is different, that is, the reference association category corresponding to the reference application is determined according to the business scenario, improving the accuracy of obtaining the reference association category corresponding to the reference application.

[0087] S104. If the target application matches the reference application, it is determined that there is an association relationship between the target application and the target business, and the reference association category is used as the target association category corresponding to the association relationship between the target application and the target business.

[0088] Specifically, the name of the target application is compared with the name of the reference application. If the name of the target application is the same as the name of the reference application, it is determined that the target application matches the reference application; here, the name of the reference application and the name of the target application can both refer to the installation package name of the application. Since the same application is released to different platforms, the name corresponding to the application may be inconsistent. Therefore, when the name of the target application and the name of the reference application both indicate the same application, it is determined that the target application matches the reference application. That is, the matching of the target application and the reference application indicates that the target application and the reference application belong to the same application. Since there is an association relationship between the reference application and the target business, it can be determined that there is an association relationship between the target application and the target business, and the reference association category is used as the target association category corresponding to the association relationship between the target application and the target business. By matching the target application with the reference application to obtain the target association category corresponding to the target application, it is not necessary to calculate the corresponding first installation proportion for the target application installed by each sample user to obtain the target association category corresponding to the target application, which can reduce the computational workload and improve the efficiency of obtaining the target association category corresponding to the target application.

[0089] For example, if the reference application is Application A, and Sample Users 1 - 3 have all installed Application A, it is only necessary to calculate the first installation ratio for Application A once, and determine the corresponding reference association category based on the first installation ratio corresponding to Application A. The applications A installed by Sample Users 1 - 3 can be matched with the reference application to obtain the target association category corresponding to the applications A installed by Sample Users 1 - 3. There is no need to calculate the corresponding first installation ratio for the applications A installed by Sample Users 1 - 3 separately, which can reduce the calculation amount and improve the efficiency of obtaining the target association category corresponding to the target application.

[0090] S105. Use the target association category and the object attribute of the sample user to adjust the user recognition model to obtain a target user recognition model for identifying target users associated with the target business.

[0091] Specifically, after obtaining the target association category, the user recognition model can be trained with the target association category and the object attribute of the sample user as features; that is, the parameters of the user recognition model are adjusted using the target association category and the object attribute of the sample user, and the adjusted user recognition model is used as the target user recognition model. The target user recognition model is used to identify target users associated with the target business. The target user can refer to a user interested in the target business; or, the target user refers to a user that fits the business characteristics of the target business; for example, in the business scenario of credit overdue, the target user refers to a user with a relatively high probability of having credit overdue characteristics. It can be seen that by determining the target association category corresponding to the association relationship between the target application and the target business according to the business scenario, and using the target association category and the object attribute as features to train the user recognition model; that is to say, dynamically and adaptively constructing the features of the user recognition model according to the business scenario, improving the accuracy of the features, and further improving the recognition accuracy of the user recognition model.

[0092] For example, such as Figure 5In this case, in the educational business scenario, the user recognition model needs to be trained using object attributes and the target association category corresponding to the target application associated with this educational business scenario; in the gaming business scenario, the user recognition model needs to be trained using object attributes and the target association category corresponding to the target application associated with this gaming business scenario. Since the target applications are inconsistent and the sample users also vary in different business scenarios, the association categories corresponding to the target applications are also inconsistent; therefore, the target association category corresponding to the target application can be referred to as a personalized feature, which is dynamically and adaptively constructed according to the business scenario. By training the user recognition model using the personalized feature and the general feature (i.e., object attributes) in the business scenario, the recognition accuracy of the user recognition model is improved.

[0093] In this application, first, the electronic device can divide the association category of the association relationship between the application programs on the market (i.e., reference application programs) and the target business according to the first installation ratio obtained by counting the first installation ratio of the reference application programs installed by the sample users, to obtain the reference association category. Then, when the target application program matches the reference application program, it is determined that there is an association relationship between the target application program and the target business, and the reference association category is used as the target association category corresponding to the association relationship between the target application program and the target business. By the method of matching the target application program with the reference application program to obtain the target association category corresponding to the target application program, it is not necessary to calculate the corresponding first installation ratio for each target application program installed by the sample users to obtain the target association category corresponding to the target application program, which can reduce the amount of computation and improve the efficiency of obtaining the target association category corresponding to the target application program. Further, the target association category and the object attributes of the sample users can be used as features to train the user recognition model to obtain the target user recognition model; in different application scenarios, the reference association categories corresponding to the reference application programs are different, so that the target association categories corresponding to the target application programs are also different. That is to say, this application can dynamically and adaptively construct the features of the user recognition model according to the business scenario, which can improve the distinguishability between application programs and thus improve the recognition accuracy of the user recognition model.

[0094] Optionally, the specific method for obtaining the reference application program having an association relationship with the target business in the above step S101 includes the following steps s11 to s14:

[0095] s11. Obtain the application installation list, which includes candidate application programs.

[0096] s12. Count the second installation ratio of the candidate application program installed by the sample users in the sample user set.

[0097] s13. Screen out candidate applications with significant features from the application list according to the second installation proportion.

[0098] s14. Use the candidate applications with significant features as reference applications having an association with the target service.

[0099] In steps s11 to s14, the electronic device can screen out candidate applications with significant features from the candidate applications on the market as reference applications. Specifically, the electronic device can obtain an application installation list, which includes candidate applications available for installation, and count the second installation proportion corresponding to the installation of the candidate applications by the sample users in the sample user set; that is, the second installation proportion is determined according to the installation proportion corresponding to the installation of the candidate applications by the positive sample users and the installation proportion corresponding to the installation of the candidate applications by the negative sample users. Furthermore, screen out candidate applications with significant features from the application list according to the second installation proportion; that is, the candidate applications with significant features refer to the applications that have a positive or negative impact on the target service. Then, the candidate applications with significant features can be used as reference applications having an association with the target service; subsequently, only the reference applications with significant features need to be analyzed, without analyzing all the applications on the market, which can improve the efficiency of analyzing applications and save costs.

[0100] Optionally, step s12 above may include the following step s21 or step s22:

[0101] s21. Use the candidate applications in the application installation list with the second installation proportion greater than the first installation proportion threshold as candidate applications with significant features; or,

[0102] s23. Use the candidate applications in the application installation list with the second installation proportion less than the second installation proportion threshold as candidate applications with significant features; the first installation proportion threshold is greater than the second installation proportion threshold.

[0103] In steps S21 and S22, when the second installation ratio corresponding to the candidate application is greater than the first installation ratio, it indicates that positive sample users prefer to install the candidate application, that is, the candidate application has a positive impact on the target service; when the second installation ratio corresponding to the candidate application is less than the second installation ratio, it indicates that negative sample users prefer to install the candidate application, that is, the candidate application has a negative impact on the target service. When the second installation ratio corresponding to the candidate application is greater than the second installation ratio and less than the first installation ratio, it indicates that both positive sample users and negative sample users like to install the candidate application, that is, the impact of the candidate application on the target service is relatively small or has no impact. Therefore, the electronic device can use the candidate application with a second installation ratio greater than the first installation ratio threshold in the application installation list as a candidate application with significant features; or, it can use the candidate application with a second installation ratio less than the second installation ratio threshold in the application installation list as a candidate application with significant features. The candidate applications with a second installation ratio greater than or equal to the second installation ratio and less than or equal to the first installation ratio can be filtered out, and subsequent analysis of candidate applications without significance is not required, which can improve the efficiency of analyzing applications and save costs.

[0104] Optionally, the sample users in the sample user set include positive sample users and negative sample users. The positive sample users are the sample users in the sample user set marked as associated with the target service, and the negative sample users are the sample users in the sample user set marked as not related to the target service;

[0105] Optionally, the above step S102 may include the following steps S31 to S32:

[0106] S31. Statistically calculate the first installation share of the reference application installed by positive sample users in the sample user set, and the second installation share of the reference application installed by negative sample users in the sample user set.

[0107] S32. Determine the first installation ratio according to the first installation share and the second installation share.

[0108] In steps S31 to S32, the electronic device can use the installation ratio of the reference application among positive sample users as the first installation share of positive sample users installing the reference application; and use the installation ratio of the reference application among negative sample users as the second installation share of negative sample users installing the reference application. Then, the first installation ratio can be determined according to the first installation share and the second installation share; among them, the specific method for calculating the first installation ratio can refer to the above formula (3), which will not be elaborated here.

[0109] Optionally, step s31 described above may include the following steps s41 to s43:

[0110] s41. Count the number of positive sample users, the number of negative sample users, the second installation volume of the reference application installed by the positive sample users, and the third installation volume of the reference application installed by the negative sample users in the sample user set.

[0111] s42. Take the ratio between the second installation volume and the number of positive sample users as the first installation share.

[0112] s43. Take the ratio between the third installation volume and the number of negative sample users as the second installation share.

[0113] In steps s41 to s43, the larger the first installation share, the more users in the positive sample users install the reference application; that is, the positive sample users prefer to install the reference application. The smaller the first installation share, the fewer users in the positive sample users install the reference application; that is, the positive sample users do not prefer to install the reference application. Similarly, the larger the second installation share, the more users in the negative sample users install the reference application; that is, the negative sample users prefer to install the reference application. The smaller the second installation share, the fewer users in the negative sample users install the reference application; that is, the negative sample users do not prefer to install the reference application.

[0114] Optionally, step S103 described above may include the following steps s51 to s54:

[0115] s51. Obtain an application category tree; the application category tree includes a root node, a first leaf node, and a second leaf node connected to the root node.

[0116] s52. Add the reference application to the root node of the application category tree.

[0117] s53. If the first installation ratio is greater than the first installation ratio threshold, add the reference application to the first leaf node and determine that the reference association category is positively correlated.

[0118] s54. If the first installation ratio is less than the second installation ratio threshold, add the reference application to the second leaf node and determine that the reference association category is negatively correlated; the first installation ratio threshold is greater than the second installation ratio threshold.

[0119] In steps S51 - S54, the electronic device can add a reference application to the application category tree; the application category tree includes a root node, a first leaf node and a second leaf node connected to the root node. The root node is used to store reference applications having an association relationship with the target service. The first leaf node is used to store reference applications in the root node where the target association category is positively correlated. The second leaf node is used to store reference applications in the root node where the target association category is negatively correlated. Therefore, the reference application can be added to the root node. If the first installation ratio is greater than the first installation ratio threshold, it indicates that positive sample users are more likely to install the reference application, that is, the reference application has a positive impact on the target service, then the reference application is added to the first leaf node, and the reference association category is determined to be positively correlated. If the first installation ratio is less than the second installation ratio threshold, it indicates that negative sample users are more likely to install the reference application, that is, the reference application has a negative impact on the target service, then the reference application is added to the second leaf node, and the reference association category is determined to be negatively correlated. By establishing the application category tree, it is beneficial to quickly query the target association category corresponding to the target application and improve the efficiency of determining the target association category of the target application.

[0120] In this embodiment, the above step S104 may include the following steps S61 - S64:

[0121] S61. Traverse the root node of the application category tree. If the target application matches the reference application in the root node, it is determined that there is an association relationship between the target application and the target service, and traverse the first leaf node according to the node path of the application category tree.

[0122] S62. If the target application matches the reference application in the first leaf node, it is determined that the target association category is positively correlated.

[0123] S63. If the target application does not match the reference application in the first leaf node, traverse the second leaf node according to the node path of the application category tree.

[0124] S64. If the target application matches the reference application in the second leaf node, it is determined that the target association category is negatively correlated.

[0125] In steps S61 - S64, as Figure 6As shown, the electronic device can query the target association category corresponding to the association relationship between the target application and the target service through the application category tree. Specifically, the electronic device can traverse the root node of the application category tree. If the target application does not match the reference application in the root node, it indicates that there is no association relationship between the target application and the target service, that is, the target application does not have significance, and this process can be ended. If the target application matches the reference application in the root node, it indicates that the target application has significant features, then it is determined that there is an association relationship between the target application and the target service. Further, according to the node path of the application category tree, the first leaf node can be traversed. If the target application matches a certain reference application in the first leaf node, it is determined that the target association category is positively correlated; if the target application does not match the reference application in the first leaf node, it indicates that the target application does not belong to the first leaf node, then according to the node path of the application category tree, the second leaf node can be traversed. If the target application matches the reference application in the second leaf node, it is determined that the target association category is negatively correlated. The node path of the application category tree can be from top to bottom, from left to right, and of course, it can also be other ways, which are not limited here. By querying the application category tree, the target association category corresponding to the target application is determined, which improves the efficiency of determining the target association category corresponding to the target application, reduces the amount of computation, and reduces costs.

[0126] Optionally, the positive correlation includes a first-level positive correlation and a second-level positive correlation, and the application category tree further includes a third leaf node and a fourth leaf node connected to the first leaf node;

[0127] The above step s53 may include the following steps s71 to s76:

[0128] s71. If the first installation ratio is greater than the first installation ratio threshold, add the reference application to the first leaf node, and obtain the first installation amount of the reference application installed by the sample user.

[0129] s72. Determine the coverage rate of the reference application according to the first installation amount, and sort the reference application according to the first installation ratio.

[0130] s73. Cumulatively calculate the coverage rate of the reference application in sequence according to the sorting order to obtain the sum of the coverage rates.

[0131] s74. Group the reference applications according to the sum of the coverage rates to obtain a first group and a second group.

[0132] s75. If the reference application belongs to the first group, add the reference application to the third leaf node, and determine that the reference association category is positively correlated at the first level.

[0133] s76. If the reference application belongs to the second group, add the reference application to the fourth leaf node, determine that the reference association category is positively correlated at the second level, and the first installation proportion corresponding to the reference application in the first group is greater than the first installation proportion corresponding to the reference application in the second group.

[0134] In steps s71 to s76, as Figure 2a shown, the positive correlation includes positive correlation at the first level and positive correlation at the second level. The application category tree further includes a third leaf node and a fourth leaf node connected to the first leaf node; the third leaf node is used to store reference applications whose association relationship with the target service in the first leaf node belongs to the positively correlated category at the first level, and the fourth leaf node is used to store reference applications whose association relationship with the target service in the first leaf node belongs to the positively correlated category at the second level. Among them, the degree of association between the reference application belonging to the positively correlated category at the first level and the target service is greater than the degree of association between the reference application belonging to the positively correlated category at the second level and the target service. Therefore, if the first installation proportion is greater than the first installation proportion threshold, indicating that the reference application has a positive impact on the target service, add the reference application to the first leaf node, and obtain the first installation quantity of the reference application installed by the sample user. Take the ratio between the first installation quantity and the total number of positive sample users as the coverage rate of the reference application, and sort the reference applications in descending order of the first installation proportion, or in ascending order. Further, accumulate the coverage rates of the reference applications in sequence according to the sorting order to obtain the sum of the coverage rates. Divide the reference applications into at least two groups according to the sum of the coverage rates. Here, take the groups including the first group and the second group as an example. The first installation proportion corresponding to each reference application in the first group is greater than the first installation proportion corresponding to the reference application in the second group; and the sum of the coverage rates corresponding to the applications in the first group and the second group is greater than the first coverage rate threshold and less than the second coverage rate threshold. If the reference application belongs to the first group, add the reference application to the third leaf node, and determine that the reference association category is positively correlated at the first level; if the reference application belongs to the second group, add the reference application to the fourth leaf node, and determine that the reference association category is positively correlated at the second level. By further subdividing the positive correlation corresponding to the reference application according to the coverage rate, more refined information can be provided for training the user recognition model, and the recognition accuracy of the user recognition model can be improved; the discrimination of the application can be ensured, and the coverage rate of the reference application can be ensured.

[0135] Optionally, the above step s62 may include the following steps s81 to s84:

[0136] s81. If the target application matches the reference application in the first leaf node, traverse the third leaf node according to the node path of the application category tree.

[0137] s82. If the target application matches the reference application in the third leaf node, determine that the target association category is positively correlated at the first level.

[0138] s83. If the target application does not match the reference application in the third leaf node, traverse the fourth leaf node according to the node path of the application category tree.

[0139] s84. If the target application matches the reference application in the fourth leaf node, determine that the target association category is positively correlated at the second level.

[0140] In the above steps s81 to s84, as Figure 7 In, when the target association category belongs to positive correlation, the electronic device can further determine whether the target association category belongs to positive correlation at the first level or positive correlation at the second level. Specifically, if the target application matches the reference application in the first leaf node, it indicates that the target association category between the target application and the target service belongs to positive correlation, and then traverse the third leaf node according to the node path of the application category tree. If the target application matches the reference application in the third leaf node, determine that the target association category is positively correlated at the first level. If the target application does not match the reference application in the third leaf node, traverse the fourth leaf node according to the node path of the application category tree. If the target application matches the reference application in the fourth leaf node, determine that the target association category is positively correlated at the second level. Through the application category tree, the level to which the positive correlation corresponding to the target application belongs can be queried, that is, the target application can be subdivided into a target application belonging to positive correlation at the first level and a target application belonging to positive correlation at the second level, which can provide more detailed feature information for training the user recognition model and improve the recognition accuracy of the user recognition model.

[0141] Optionally, the above step S105 may include the following steps s91 to s94:

[0142] s91. Count the number of applications with the target association category among the target applications installed by the sample user.

[0143] S92. Obtain the labeled associated label of the sample user, where the labeled associated label is used to reflect whether there is an association between the sample user and the target service.

[0144] S93. Use the user identification model to perform an association identification on the number of application programs and the object attributes of the sample user to obtain a predicted associated label.

[0145] S94. Adjust the user identification model according to the labeled associated label and the predicted associated label to obtain the target user identification model.

[0146] In steps S91 - S94, as Figure 8 In this case, the electronic device can count the number of application programs with the target association category among the target application programs installed by the sample user, and obtain the labeled associated label of the sample user. The labeled associated label of the sample user is manually labeled and is used to reflect whether there is an association between the sample user and the target service. Further, the user identification model can be used to perform an association identification on the number of application programs and the object attributes of the sample user to obtain a predicted associated label. If the labeled associated label is close to the predicted associated label, it indicates that the recognition error of the user identification model is relatively low. On the contrary, if the labeled associated label is quite different from the predicted associated label, it indicates that the recognition error of the user identification model is relatively high. Therefore, the recognition error of the user identification model can be calculated based on the labeled associated label and the predicted associated label. When the recognition error is less than the error threshold, it indicates that the accuracy of the user identification model is relatively high, and the user identification model is used as the target user identification model. When the recognition error is greater than or equal to the error threshold, it indicates that the accuracy of the user identification model is relatively low. The user identification model is adjusted according to the recognition error. When the error of the adjusted user identification model is greater than the error threshold, the adjusted user identification model is used as the target user identification model. By training the user identification model according to the number of target application programs belonging to the target association category installed by the sample user and the object attributes, the recognition accuracy of the user identification model can be improved.

[0147] Optionally, the method may further include the following steps S111 - S113:

[0148] S111. Receive an explanation request for the predicted associated label; extract business keywords associated with the target service according to the explanation request.

[0149] S112. Compare the business keywords with the target application programs to determine the business relationship between the target application programs and the target service.

[0150] S113. Generate explanation information using the business relationship, where the explanation information is used to explain the influencing factor of the business relationship on the predicted associated label.

[0151] In steps S111 to S113, the electronic device can determine the service relationship between the target application and the target service according to the service keyword. Specifically, the electronic device can receive an explanation request for the predicted association label, and extract the service keyword associated with the target service according to the explanation request; the service keyword of the target service can refer to the fields included in the target service name, or the fields in the service name corresponding to the target service. Further, compare the service keyword with the target application to determine the service relationship between the target application and the target service. The service relationship between the target application and the target service includes the same category relationship and the non-same category relationship. The same category relationship is used to indicate that the target application and the target service belong to the same service scenario, and the non-same category relationship is used to indicate that the target application and the target service do not belong to the same service scenario. For example, in the target service of pushing programming network courses, the target service keyword includes training. If the target application is a training application, it indicates that the target application and the push programming network courses both belong to the education scenario. Generally, target applications belonging to the same category also have a positive impact on the target service, and target applications belonging to different categories have a negative impact on the target service. Therefore, the service relationship is used to generate explanation information, and the explanation information is used to explain the influence factor of the service relationship on the predicted association label. Through the explanation information, the user can trace the reason for the user recognition model to output the predicted association label, and improve the credibility of the user recognition model.

[0152] Optionally, the above step S112 may include the following steps S121 to S122:

[0153] S121. If the target application includes the service keyword, determine that the service relationship between the target application and the target service is the same category relationship.

[0154] S122. If the target application does not include the service keyword, determine that the service relationship between the target application and the target service is the non-same category relationship.

[0155] In steps S121 to S122, the electronic device can compare the name of the target application with the service keyword. If the target application (i.e., the name of the target application) includes the service keyword, determine that the service relationship between the target application and the target service is the same category relationship; if the target application does not include the service keyword, determine that the service relationship between the target application and the target service is the non-same category relationship.

[0156] Optionally, the method may include the following steps S131 to S133:

[0157] s131. Receive an identification request for a target user, where the identification request includes the object attributes of the target user and the historical applications installed by the target user in the past.

[0158] s132. If the historical application matches the reference application, use the reference association category as the historical association category between the historical application and the target service.

[0159] s133. Use the target user identification model to perform relevance category identification on the historical association category and the object attributes of the target user, and obtain a target association label indicating whether there is an association between the target user and the target service.

[0160] In steps s131 to s133, after training the target user identification model, the target user identification model can be used to expand users for the target service. Specifically, the electronic device can receive an identification request for a target user, where the identification request includes the object attributes of the target user and the historical applications installed by the target user in the past. Compare the historical applications with the reference applications. If the historical application does not match the reference application, it is determined that the historical application does not have significant features, and the historical application is filtered out, that is, the historical application is not processed. If the historical application matches the reference application, it is determined that the historical application has significant features, and the reference association category is used as the historical association category between the historical application and the target service. Further, use the target user identification model to perform relevance category identification on the historical association category and the object attributes of the target user, and obtain a target association label indicating whether there is an association between the target user and the target service.

[0161] Please refer to Figure 9 , which is a schematic structural diagram of a user identification device provided by an embodiment of the present application. The above user identification device can be a computer program (including program code) running in a computer device. For example, the user identification device is an application software; the device can be used to execute the corresponding steps in the method provided by an embodiment of the present application. As Figure 9 shown, the user identification device may include: an acquisition module 901, a division module 902, a determination module 903, an adjustment module 904, an explanation module 905, and a user identification module 906.

[0162] The acquisition module 901 is used to acquire a sample user set for a target service, the target applications installed by the sample users in the sample user set, the object attributes of the sample users, and the reference applications having an association relationship with the target service;

[0163] A division module 902, which is used to count the first installation ratio corresponding to the installation of the reference application by the sample users in the sample user set, and perform an association category division on the association relationship between the reference application and the target service according to the first installation ratio to obtain a reference association category;

[0164] A determination module 903, which is used to determine that there is an association relationship between the target application and the target service if the target application matches the reference application, and use the reference association category as the target association category corresponding to the association relationship between the target application and the target service;

[0165] An adjustment module 904, which is used to adjust the user recognition model by using the target association category and the object attributes of the sample user to obtain a target user recognition model for recognizing target users associated with the target service.

[0166] Optionally, the division model performs an association category division on the association relationship between the reference application and the target service according to the first installation ratio to obtain a reference association category, including:

[0167] Obtain an application category tree; the application category tree includes a root node, as well as a first leaf node and a second leaf node connected to the root node;

[0168] Add the reference application to the root node of the application category tree;

[0169] If the first installation ratio is greater than the first installation ratio threshold, add the reference application to the first leaf node and determine that the reference association category is positively correlated;

[0170] If the first installation ratio is less than the second installation ratio threshold, add the reference application to the second leaf node and determine that the reference association category is negatively correlated; the first installation ratio threshold is greater than the second installation ratio threshold.

[0171] Optionally, if the target application matches the reference application, the determination module determines that there is an association relationship between the target application and the target service, and uses the reference association category as the target association category corresponding to the association relationship between the target application and the target service, including:

[0172] Traverse the root node of the application category tree. If the target application matches the reference application in the root node, determine that there is an association relationship between the target application and the target service, and traverse the first leaf node according to the node path of the application category tree;

[0173] If the target application matches the reference application in the first leaf node, determine that the target association category is positively correlated;

[0174] If the target application does not match the reference application in the first leaf node, traverse the second leaf node according to the node path of the application category tree;

[0175] If the target application matches the reference application in the second leaf node, determine that the target association category is negatively correlated.

[0176] Optionally, the positive correlation includes a first-level positive correlation and a second-level positive correlation, and the application category tree further includes a third leaf node and a fourth leaf node connected to the first leaf node;

[0177] If the first installation ratio is greater than the first installation ratio threshold, the partitioning module adds the reference application to the first leaf node and determines that the reference association category is positively correlated; including:

[0178] If the first installation ratio is greater than the first installation ratio threshold, add the reference application to the first leaf node and obtain the first installation quantity of the reference application installed by the sample user;

[0179] Determine the coverage rate of the reference application according to the first installation quantity, and sort the reference application according to the first installation ratio;

[0180] Accumulate the coverage rates of the reference applications in sequence according to the sorting order to obtain the sum of the coverage rates;

[0181] Group the reference applications according to the sum of the coverage rates to obtain a first group and a second group;

[0182] If the reference application belongs to the first group, add the reference application to the third leaf node and determine that the reference association category is the first-level positive correlation;

[0183] If the reference application belongs to the second group, add the reference application to the fourth leaf node and determine that the reference association category is the second-level positive correlation, and the first installation ratio corresponding to the reference application in the first group is greater than the first installation ratio corresponding to the reference application in the second group.

[0184] Optionally, if the target application matches the reference application in the first leaf node, the determining module determines that the target association category is positively correlated; including:

[0185] If the target application matches the reference application in the first leaf node, traverse the third leaf node according to the node path of the application category tree;

[0186] If the target application matches the reference application in the third leaf node, determine that the target associated category is positively correlated at the first level;

[0187] If the target application does not match the reference application in the third leaf node, traverse the fourth leaf node according to the node path of the application category tree;

[0188] If the target application matches the reference application in the fourth leaf node, determine that the target associated category is positively correlated at the second level.

[0189] Optionally, the adjustment module adjusts the user recognition model by using the target associated category and the object attributes of the sample user to obtain a target user recognition model for recognizing a target user associated with the target service, including:

[0190] Count the number of applications with the target associated category among the target applications installed by the sample user;

[0191] Obtain the labeled associated label of the sample user, and the labeled associated label is used to reflect whether the sample user has an associated relationship with the target service;

[0192] Use the user recognition model to perform an association recognition on the number of applications and the object attributes of the sample user to obtain a predicted associated label;

[0193] Adjust the user recognition model according to the labeled associated label and the predicted associated label to obtain the target user recognition model.

[0194] Optionally, the user recognition device further includes an explanation module, and the explanation module is configured to receive an explanation request for the predicted associated label; extract business keywords associated with the target service according to the explanation request;

[0195] Compare the business keywords with the target application to determine the business relationship between the target application and the target service;

[0196] Generate explanation information by using the business relationship, and the explanation information is used to explain the influence factor of the business relationship on the predicted associated label.

[0197] Optionally, the explanation module compares the business keywords with the target application to determine the business relationship between the target application and the target service, including:

[0198] If the target application includes the business keyword, determine that the business relationship between the target application and the target business is a same-category relationship;

[0199] If the target application does not include the business keyword, determine that the business relationship between the target application and the target business is a non-same-category relationship.

[0200] Optionally, the obtaining module obtains a reference application having an association relationship with the target business, including:

[0201] Obtain an application installation list, where the application installation list includes candidate applications;

[0202] Count a second installation ratio corresponding to the installation of the candidate application by the sample users in the sample user set;

[0203] Filter out candidate applications with significant features from the application list according to the second installation ratio;

[0204] Use the candidate applications with significant features as the reference applications having an association relationship with the target business.

[0205] Optionally, the obtaining module filters out candidate applications with significant features from the application list according to the second installation ratio, including:

[0206] Use the candidate applications in the application installation list with a second installation ratio greater than a first installation ratio threshold as the candidate applications with significant features; or,

[0207] Use the candidate applications in the application installation list with a second installation ratio less than a second installation ratio threshold as the candidate applications with significant features; the first installation ratio threshold is greater than the second installation ratio threshold.

[0208] Optionally, the sample users in the sample user set include positive sample users and negative sample users. The positive sample users are the sample users in the sample user set labeled as associated with the target business, and the negative sample users are the sample users in the sample user set labeled as not related to the target business;

[0209] The partitioning module counts a first installation ratio corresponding to the installation of the reference application by the sample users in the sample user set, including:

[0210] Count a first installation share of the positive sample users in the sample user set installing the reference application, and a second installation share of the negative sample users in the sample user set installing the reference application;

[0211] Determine the first installation proportion based on the first installation share and the second installation share.

[0212] Optionally, the partitioning module counts the first installation share of the positive sample users in the sample user set who install the reference application, and the second installation share of the negative sample users in the sample user set who install the reference application, including:

[0213] Count the number of positive sample users, the number of negative sample users, the second installation quantity of the positive sample users who install the reference application, and the third installation quantity of the negative sample users who install the reference application in the sample user set;

[0214] Use the ratio between the second installation quantity and the number of positive sample users as the first installation share;

[0215] Use the ratio between the third installation quantity and the number of negative sample users as the second installation share.

[0216] Optionally, the user identification device further includes a user identification module, and the user identification module is configured to: receive an identification request for a target user, where the identification request includes the object attribute of the target user and the historical applications installed by the target user in the past;

[0217] If the historical application matches the reference application, use the reference association category as the historical association category between the historical application and the target service;

[0218] Use the target user identification model to perform an association category identification on the historical association category and the object attribute of the target user, and obtain a target association label for indicating whether there is an association between the target user and the target service.

[0219] According to an embodiment of the present application, Figure 4 The steps involved in the user identification method shown can be Figure 9 executed by each module in the user identification device shown. For example, Figure 4 The step S101 shown in Figure 9 can be executed by the acquisition module 901 in Figure 4 The steps S102 and S103 shown in Figure 9 can be executed by the partitioning module 902 in Figure 4 The step S104 shown in Figure 9 can be executed by the determination module 905 in Figure 4 The step S105 shown in Figure 9 can be executed by the adjustment module 905 in

[0220] According to an embodiment of the present application, Figure 9Each module in the user identification device shown can be separately or wholly combined into one or several units to form, or a certain one (or some) of the units can be further split into multiple smaller sub-units in terms of function, and the same operations can be achieved without affecting the realization of the technical effects of the embodiments of this application. The above-mentioned modules are divided based on logical functions. In practical applications, the function of one module can also be realized by multiple units, or the functions of multiple modules can be realized by one unit. In other embodiments of this application, the user identification device can also include other units. In practical applications, these functions can also be assisted by other units and can be realized through the cooperation of multiple units.

[0221] According to an embodiment of this application, it can be achieved by running a computer program (including program code) that can execute the respective steps involved in the corresponding method shown on a general computer device such as a computer including processing elements and storage elements such as a central processing unit (CPU), a random access storage medium (RAM), and a read-only storage medium (ROM). Figure 4 to construct the user identification device shown in Figure 9 and to implement the user identification method of the embodiments of this application. The above-mentioned computer program can be recorded on a computer-readable recording medium, for example, and loaded into the above-mentioned computing device through the computer-readable recording medium and run therein.

[0222] In this application, first, the electronic device can calculate the first installation ratio corresponding to the reference application installed by the sample user, and classify the association categories of the association relationship between the application programs (i.e., reference application programs) on the market and the target service according to the first installation ratio to obtain the reference association categories. Then, when the target application program matches the reference application program, it is determined that there is an association relationship between the target application program and the target service, and the reference association category is used as the target association category corresponding to the association relationship between the target application program and the target service. By the method of matching the target application program with the reference application program to obtain the target association category corresponding to the target application program, it is not necessary to calculate the corresponding first installation ratio for each target application program installed by the sample user to obtain the target association category corresponding to the target application program, which can reduce the amount of calculation and improve the efficiency of obtaining the target association category corresponding to the target application program. Further, the target association category and the object attributes of the sample user can be used as features to train the user identification model to obtain the target user identification model; in different application scenarios, the reference association category corresponding to the reference application program is different, so that the target association category corresponding to the target application program is also different. That is to say, this application can dynamically and adaptively construct the features of the user identification model according to the business scenario, which can improve the discrimination between application programs and thus improve the identification accuracy of the user identification model.

[0223] Please refer toFigure 10 , which is a schematic structural diagram of a computer device provided by an embodiment of the present application. As Figure 10 shown, the above computer device 1000 may include: a processor 1001, a network interface 1004, and a memory 1005. In addition, the above computer device 1000 may further include: a user interface 1003 and at least one communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. Among them, the user interface 1003 may include a display screen (Display) and a keyboard (Keyboard). Optionally, the user interface 1003 may further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. Optionally, the memory 1005 may further be at least one storage device located far from the foregoing processor 1001. As Figure 10 shown, the memory 1005, as a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a device control application program.

[0224] In Figure 10 the computer device 1000 shown, the network interface 1004 can provide network communication functions; while the user interface 1003 is mainly used to provide an input interface for users; and the processor 1001 can be used to call the device control application program stored in the memory 1005 to implement:

[0225] Obtain a sample user set for a target service, target applications installed by sample users in the sample user set, object attributes of the sample users, and reference applications having an association relationship with the target service;

[0226] Statistically calculate a first installation ratio corresponding to the installation of the reference application by sample users in the sample user set, and perform an association category division on the association relationship between the reference application and the target service according to the first installation ratio to obtain a reference association category;

[0227] If the target application matches the reference application, it is determined that there is an association relationship between the target application and the target service, and the reference association category is used as the target association category corresponding to the association relationship between the target application and the target service;

[0228] Adjust the user recognition model by using the target association category and the object attributes of the sample users to obtain a target user recognition model for identifying target users associated with the target service.

[0229] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to implement classifying the association relationship between the reference application and the target service according to the first installation ratio to obtain a reference association class, including:

[0230] Obtain an application category tree; the application category tree includes a root node, as well as a first leaf node and a second leaf node connected to the root node;

[0231] Add the reference application to the root node of the application category tree;

[0232] If the first installation ratio is greater than the first installation ratio threshold, add the reference application to the first leaf node and determine that the reference association class is positively correlated;

[0233] If the first installation ratio is less than the second installation ratio threshold, add the reference application to the second leaf node and determine that the reference association class is negatively correlated; the first installation ratio threshold is greater than the second installation ratio threshold.

[0234] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to implement that if the target application matches the reference application, it is determined that there is an association relationship between the target application and the target service, and use the reference association class as the target association class corresponding to the association relationship between the target application and the target service, including:

[0235] Traverse the root node of the application category tree. If the target application matches the reference application in the root node, it is determined that there is an association relationship between the target application and the target service, and traverse the first leaf node according to the node path of the application category tree;

[0236] If the target application matches the reference application in the first leaf node, it is determined that the target association class is positively correlated;

[0237] If the target application does not match the reference application in the first leaf node, traverse the second leaf node according to the node path of the application category tree;

[0238] If the target application matches the reference application in the second leaf node, it is determined that the target association class is negatively correlated.

[0239] Optionally, the positive correlation includes a first-level positive correlation and a second-level positive correlation, and the application category tree further includes a third leaf node and a fourth leaf node connected to the first leaf node;

[0240] The processor 1001 can be used to call the device control application program stored in the memory 1005 to implement that if the first installation ratio is greater than the first installation ratio threshold, adding the reference application program to the first leaf node and determining that the reference association category is positively correlated; including:

[0241] If the first installation ratio is greater than the first installation ratio threshold, adding the reference application program to the first leaf node and obtaining the first installation quantity of the reference application program installed by the sample user;

[0242] Determining the coverage rate of the reference application program according to the first installation quantity, and sorting the reference application program according to the first installation ratio;

[0243] Accumulating the coverage rates of the reference application program in sequence according to the sorting order to obtain the sum of coverage rates;

[0244] Grouping the reference application program according to the sum of coverage rates to obtain a first group and a second group;

[0245] If the reference application program belongs to the first group, adding the reference application program to the third leaf node and determining that the reference association category is first-level positively correlated;

[0246] If the reference application program belongs to the second group, adding the reference application program to the fourth leaf node and determining that the reference association category is second-level positively correlated, and the first installation ratio corresponding to the reference application program in the first group is greater than the first installation ratio corresponding to the reference application program in the second group.

[0247] The processor 1001 can be used to call the device control application program stored in the memory 1005 to implement that if the target application program matches the reference application program in the first leaf node, determining that the target association category is positively correlated; including:

[0248] If the target application program matches the reference application program in the first leaf node, traversing the third leaf node according to the node path of the application program category tree;

[0249] If the target application program matches the reference application program in the third leaf node, determining that the target association category is first-level positively correlated;

[0250] If the target application program does not match the reference application program in the third leaf node, traversing the fourth leaf node according to the node path of the application program category tree;

[0251] If the target application program matches the reference application program in the fourth leaf node, determining that the target association category is second-level positively correlated.

[0252] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to adjust the user recognition model by using the target association category and the object attributes of the sample user, so as to obtain a target user recognition model for recognizing a target user associated with the target service, including:

[0253] Count the number of applications with the target association category among the target applications installed by the sample user;

[0254] Obtain the labeled association label of the sample user, where the labeled association label is used to reflect whether the sample user is associated with the target service;

[0255] Perform an association recognition on the number of applications and the object attributes of the sample user by using the user recognition model to obtain a predicted association label;

[0256] Adjust the user recognition model according to the labeled association label and the predicted association label to obtain the target user recognition model.

[0257] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to: receive an explanation request for the predicted association label; extract service keywords associated with the target service according to the explanation request;

[0258] Compare the service keywords with the target application to determine the service relationship between the target application and the target service;

[0259] Generate explanation information by using the service relationship, where the explanation information is used to explain the influence factor of the service relationship on the predicted association label.

[0260] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to compare the service keywords with the target application to determine the service relationship between the target application and the target service, including:

[0261] If the target application includes the service keyword, determine that the service relationship between the target application and the target service is a same-category relationship;

[0262] If the target application does not include the service keyword, determine that the service relationship between the target application and the target service is a non-same-category relationship.

[0263] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to obtain a reference application having an association relationship with the target service, including:

[0264] Obtain an application installation list, where the application installation list includes candidate applications;

[0265] Statistically calculate the second installation proportion corresponding to the installation of the candidate application by the sample users in the sample user set;

[0266] Filter out candidate applications with significant features from the application list according to the second installation proportion;

[0267] Use the candidate applications with significant features as reference applications having an association relationship with the target service.

[0268] Optionally, the processor 1001 can be used to call the device control application stored in the memory 1005 to filter out candidate applications with significant features from the application list according to the second installation proportion, including:

[0269] Use the candidate applications in the application installation list whose second installation proportion is greater than the first installation proportion threshold as candidate applications with significant features; or,

[0270] Use the candidate applications in the application installation list whose second installation proportion is less than the second installation proportion threshold as candidate applications with significant features; the first installation proportion threshold is greater than the second installation proportion threshold.

[0271] Optionally, the sample users in the sample user set include positive sample users and negative sample users. The positive sample users are the sample users in the sample user set labeled as associated with the target service, and the negative sample users are the sample users in the sample user set labeled as not associated with the target service;

[0272] The processor 1001 can be used to call the device control application stored in the memory 1005 to statistically calculate the first installation proportion corresponding to the installation of the reference application by the sample users in the sample user set, including:

[0273] Statistically calculate the first installation share of the reference application installed by the positive sample users in the sample user set and the second installation share of the reference application installed by the negative sample users in the sample user set;

[0274] Determine the first installation proportion according to the first installation share and the second installation share.

[0275] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to implement the statistics of the first installation share of the reference application installed by the positive sample users in the sample user set, and the second installation share of the reference application installed by the negative sample users in the sample user set, including:

[0276] Statistically calculate the number of positive sample users, the number of negative sample users, the second installation quantity of the reference application installed by the positive sample users, and the third installation quantity of the reference application installed by the negative sample users in the sample user set;

[0277] Use the ratio between the second installation quantity and the number of positive sample users as the first installation share;

[0278] Use the ratio between the third installation quantity and the number of negative sample users as the second installation share.

[0279] Optionally, the processor 1001 may be used to call the device control application stored in the memory 1005 to implement receiving an identification request for a target user, where the identification request includes the object attributes of the target user and the historical applications installed by the target user in the past;

[0280] If the historical application matches the reference application, use the reference association category as the historical association category between the historical application and the target service;

[0281] Use the target user identification model to perform relevance category identification on the historical association category and the object attributes of the target user, and obtain a target association label indicating whether the target user is associated with the target service.

[0282] It should be understood that the computer device 1000 described in the embodiments of the present application may execute the descriptions of the above user identification method in the corresponding embodiments of FIG. 3 and the foregoing Figure 7 and may also execute the descriptions of the above user identification device in the corresponding embodiments of the foregoing Figure 8 which will not be elaborated here. In addition, the descriptions of the beneficial effects of using the same method will not be elaborated either.

[0283] In this application, first, the electronic device can calculate the first installation ratio corresponding to the reference application installed by the sample user, and classify the association categories between the applications on the market (i.e., the reference applications) and the target business according to the first installation ratio to obtain the reference association categories. Then, when the target application matches the reference application, it is determined that there is an association relationship between the target application and the target business, and the reference association category is used as the target association category corresponding to the association relationship between the target application and the target business. By matching the target application with the reference application to obtain the target association category corresponding to the target application, it is not necessary to calculate the corresponding first installation ratio for each target application installed by the sample user to obtain the target association category corresponding to the target application, which can reduce the computational workload and improve the efficiency of obtaining the target association category corresponding to the target application. Further, the target association category and the object attributes of the sample user can be used as features to train the user recognition model to obtain the target user recognition model; in different application scenarios, the reference association categories corresponding to the reference applications are different, so that the target association categories corresponding to the target applications are also different. That is to say, this application can dynamically and adaptively construct the features of the user recognition model according to the business scenario, which can improve the distinguishability between applications and thus improve the recognition accuracy of the user recognition model.

[0284] In addition, it should be noted here that: The embodiment of the present application also provides a computer-readable storage medium, and the computer program executed by the above-mentioned user recognition device is stored in the above-mentioned computer-readable storage medium, and the above-mentioned computer program includes program instructions. When the above-mentioned processor executes the above-mentioned program instructions, it can execute the description of the above-mentioned user recognition method in the corresponding embodiment mentioned above. Therefore, it will not be repeated here. In addition, the description of the beneficial effects of using the same method will not be repeated either. For the technical details not disclosed in the embodiment of the computer-readable storage medium involved in the present application, please refer to the description of the method embodiment of the present application. Figure 4 As an example, the above-mentioned program instructions can be deployed to be executed on a computer device, or be deployed on multiple computer devices located at one place, or, be executed on multiple computer devices distributed at multiple places and interconnected through a communication network. The multiple computer devices distributed at multiple places and interconnected through a communication network can form a blockchain network.

[0285]

[0286] ​Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The above program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above various methods. Among them, the above storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0287] The above-disclosed are only the preferred embodiments of the present application. Of course, the scope of rights of the present application cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present application still fall within the scope covered by the present application.

Claims

1. A user identification method, characterized in that, Including: Obtain a set of sample users for a target business, the target applications installed by the sample users in the set of sample users, the object attributes of the sample users, and reference applications having an associated relationship with the target business; the sample users in the set of sample users include positive sample users and negative sample users, the positive sample users are the sample users in the set of sample users labeled as associated with the target business, and the negative sample users are the sample users in the set of sample users labeled as not related to the target business; Statistically calculate a first installation share of the positive sample users in the set of sample users installing the reference application, and a second installation share of the negative sample users in the set of sample users installing the reference application; determine a first installation ratio based on the first installation share and the second installation share, and obtain an application category tree; The application category tree includes a root node, as well as a first leaf node and a second leaf node connected to the root node; add the reference application to the root node of the application category tree; If the first installation ratio is greater than a first installation ratio threshold, add the reference application to the first leaf node and determine the reference association category as positively correlated; if the first installation ratio is less than a second installation ratio threshold, add the reference application to the second leaf node and determine the reference association category as negatively correlated; the first installation ratio threshold is greater than the second installation ratio threshold; If the target application matches the reference application, determine that there is an associated relationship between the target application and the target business, and use the reference association category as the target association category corresponding to the associated relationship between the target application and the target business; Adjust a user identification model using the target association category and the object attributes of the sample users to obtain a target user identification model for identifying target users associated with the target business.

2. The method according to claim 1, wherein The "If the target application matches the reference application, determine that there is an associated relationship between the target application and the target business, and use the reference association category as the target association category corresponding to the associated relationship between the target application and the target business" includes: Traverse the root node of the application category tree. If the target application matches the reference application in the root node, determine that there is an associated relationship between the target application and the target business, and traverse the first leaf node according to the node path of the application category tree; If the target application matches the reference application in the first leaf node, determine that the target association category is positively correlated; If the target application does not match the reference application in the first leaf node, traverse the second leaf node according to the node path of the application category tree; If the target application matches the reference application in the second leaf node, determine that the target association category is negatively correlated.

3. The method according to claim 2, wherein The positive correlation includes a first-level positive correlation and a second-level positive correlation. The application category tree further includes a third leaf node and a fourth leaf node connected to the first leaf node; If the first installation ratio is greater than the first installation ratio threshold, adding the reference application to the first leaf node and determining that the reference association category is positively correlated; includes: If the first installation ratio is greater than the first installation ratio threshold, adding the reference application to the first leaf node and obtaining the first installation quantity of the reference application installed by the sample user; Determining the coverage rate of the reference application according to the first installation quantity, and sorting the reference applications according to the first installation ratio; Accumulating the coverage rates of the reference applications in sequence according to the sorting order to obtain the sum of coverage rates; Grouping the reference applications according to the sum of coverage rates to obtain a first group and a second group; If the reference application belongs to the first group, adding the reference application to the third leaf node and determining that the reference association category is the first-level positive correlation; If the reference application belongs to the second group, adding the reference application to the fourth leaf node and determining that the reference association category is the second-level positive correlation. The first installation ratio corresponding to the reference applications in the first group is greater than the first installation ratio corresponding to the reference applications in the second group.

4. The method according to claim 3, wherein If the target application matches the reference application in the first leaf node, determining that the target association category is positively correlated; includes: If the target application matches the reference application in the first leaf node, traversing the third leaf node according to the node path of the application category tree; If the target application matches the reference application in the third leaf node, determining that the target association category is the first-level positive correlation; If the target application does not match the reference application in the third leaf node, traversing the fourth leaf node according to the node path of the application category tree; If the target application matches the reference application in the fourth leaf node, determining that the target association category is the second-level positive correlation.

5. The method according to any one of claims 1 to 4, characterized in that Adjusting the user recognition model by using the target association category and the object attribute of the sample user to obtain a target user recognition model for recognizing target users associated with the target service, includes: Counting the number of applications with the target association category among the target applications installed by the sample user; Obtaining the labeled association label of the sample user, where the labeled association label is used to reflect whether the sample user has an association relationship with the target service; Performing an association recognition on the application quantity and the object attribute of the sample user by using the user recognition model to obtain a predicted association label; Adjusting the user recognition model according to the labeled association label and the predicted association label to obtain the target user recognition model.

6. The method according to claim 5, wherein The method further includes: Receive an explanation request for the predicted association label; extract business keywords associated with the target business according to the explanation request; Compare the business keywords with the target application to determine the business relationship between the target application and the target business; Generate explanation information using the business relationship, where the explanation information is used to explain the influencing factor of the business relationship on the predicted association label.

7. The method according to claim 6, wherein The comparing the business keywords with the target application to determine the business relationship between the target application and the target business includes: If the target application includes the business keywords, determine that the business relationship between the target application and the target business is a same-category relationship; If the target application does not include the business keywords, determine that the business relationship between the target application and the target business is a non-same-category relationship.

8. The method according to claim 1, characterized in that The obtaining a reference application having an association relationship with the target business includes: Obtain an application installation list, where the application installation list includes candidate applications; Count the second installation ratio corresponding to the installation of the candidate applications by the sample users in the sample user set; Filter out candidate applications with significant features from the application list according to the second installation ratio; Use the candidate applications with significant features as reference applications having an association relationship with the target business.

9. The method according to claim 8, characterized in that, The filtering out candidate applications with significant features from the application list according to the second installation ratio includes: Use the candidate applications in the application installation list with a second installation ratio greater than the first installation ratio threshold as candidate applications with significant features; or, Use the candidate applications in the application installation list with a second installation ratio less than the second installation ratio threshold as candidate applications with significant features; the first installation ratio threshold is greater than the second installation ratio threshold.

10. The method according to claim 1, characterized in that, The method further includes: Receive an identification request for a target user, where the identification request includes the object attribute of the target user and the historical applications installed by the target user in the past; If the historical applications match the reference applications, use the reference association category as the historical association category between the historical applications and the target business; Use the target user identification model to perform an association category identification on the historical association category and the object attribute of the target user to obtain a target association label for indicating whether there is an association between the target user and the target business.

11. A user identification device, characterized in that, Includes: An acquisition module, configured to acquire a set of sample users for a target service, target applications installed by the sample users in the set of sample users, object attributes of the sample users, and reference applications having an association relationship with the target service; the sample users in the set of sample users include positive sample users and negative sample users, the positive sample users are sample users in the set of sample users labeled as being associated with the target service, and the negative sample users are sample users in the set of sample users labeled as not being related to the target service; A division module, configured to count a first installation share of the positive sample users in the set of sample users installing the reference application, and a second installation share of the negative sample users in the set of sample users installing the reference application; determine a first installation ratio according to the first installation share and the second installation share, and obtain an application category tree; The application category tree includes a root node, and a first leaf node and a second leaf node connected to the root node; add the reference application to the root node of the application category tree; If the first installation ratio is greater than a first installation ratio threshold, add the reference application to the first leaf node and determine that the reference association category is positively correlated; if the first installation ratio is less than a second installation ratio threshold, add the reference application to the second leaf node and determine that the reference association category is negatively correlated; the first installation ratio threshold is greater than the second installation ratio threshold; A determination module, configured to determine that there is an association relationship between the target application and the target service if the target application matches the reference application, and use the reference association category as the target association category corresponding to the association relationship between the target application and the target service; An adjustment module, configured to adjust a user identification model by using the target association category and the object attributes of the sample users to obtain a target user identification model for identifying target users associated with the target service.

12. A computer device, characterized in that, Comprising: A processor and a memory; The processor is connected to the memory, wherein the memory is used to store program code, and the processor is used to call the program code to execute the method according to any one of claims 1-10.

13. A computer-readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, and the computer program is adapted to be loaded and executed by the processor to execute the method according to any one of claims 1-10.

Images