Desensitized user behavior feature analysis method, device, server and medium

Abstract

The present invention provides a desensitized user behavior feature analysis method, device, server and medium, including: obtaining a plurality of user interface UI atlases collected by a user end from a target application platform, and preprocessing the interface UI atlases; generating a text data set corresponding to the interface UI atlases through an image recognition model; determining a desensitized user behavior feature vector corresponding to each user based on the text data set through a text analysis big model, the desensitized user behavior feature vector including the user's data scores for a plurality of behavior feature dimensions, and the behavior feature dimensions are determined through a natural language big model; analyzing the behavior correlation between any two users based on the desensitized user behavior feature vector corresponding to each user, and obtaining a behavior correlation analysis result. The present invention can effectively protect user privacy in the data mining process, and can also effectively improve the quality of the behavior feature dimensions, thereby helping to improve the accuracy of user behavior feature analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of computer data mining and data analysis, and in particular to a method, device, server and medium for analyzing desensitized user behavior characteristics. Background Art

[0002] Data mining is a process that involves extracting valuable information and patterns from large amounts of data. In the field of data mining, common problems include data quality, feature selection, and overfitting. In large-scale data sets, it is often ambiguous to determine the characteristics of a piece of data. This is because high-dimensional data is often accompanied by dimensionality disasters, data noise, high correlation between features, and computing resource limitations. In addition, data privacy and security issues are also an important challenge. Some data analysis platforms obtain some user data through data crawling, interface access, etc. These data are often not desensitized data, so personal privacy protection becomes increasingly difficult in the era of big data. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a desensitized user behavior feature analysis method, device, server and medium, which can effectively protect user privacy during the data mining process, and can also effectively improve the quality of the behavior feature dimension, thereby helping to improve the accuracy of user behavior feature analysis.

[0004] In a first aspect, the present invention provides a method for analyzing desensitized user behavior characteristics, the method is applied to a server, the server is connected to multiple user terminals for communication, and the method includes:

[0005] Obtaining UI atlases of multiple users collected by the user end from the target application platform, and preprocessing the UI atlases;

[0006] Generate a text dataset corresponding to the preprocessed UI atlas through an image recognition model. The text dataset is used to describe the content displayed by the UI atlas in text form.

[0007] Through the text analysis big model, the desensitized user behavior feature vector corresponding to each user is determined based on the text data set. The desensitized user behavior feature vector includes the user's data scores for multiple behavior feature dimensions. The behavior feature dimensions are determined by the natural language big model.

[0008] Based on the desensitized user behavior feature vector corresponding to each user, the behavior correlation between any two users is analyzed to obtain the behavior correlation analysis result.

[0009] In one embodiment, before determining the desensitized user behavior feature vector corresponding to each user based on the text data set by using the text analysis big model, the method further includes:

[0010] Through the natural language big model, the behavioral feature dimensions are divided based on the pre-configured user description information to obtain multiple behavioral feature dimensions corresponding to the user.

[0011] In one implementation, the text analysis model uses a deep learning model based on the Transform architecture; through the text analysis model, a desensitized user behavior feature vector corresponding to each user is determined based on the text data set, including:

[0012] For each user, a large text analysis model is used to capture the long-distance dependencies in the text dataset corresponding to the user using the self-attention mechanism and position encoding. Based on the long-distance dependencies, the data scores of the user for multiple behavioral feature dimensions are determined to construct a desensitized user behavior feature vector corresponding to the user.

[0013] In one implementation, based on the desensitized user behavior feature vector corresponding to each user, the behavior correlation between any two users is analyzed to obtain a behavior correlation analysis result, including:

[0014] For any two users, the Pearson correlation coefficient between the first user and the second user is determined based on the data scores corresponding to each behavior feature dimension in the desensitized user behavior feature vectors corresponding to the first user and the second user respectively, and the Pearson correlation coefficient is used as the behavior correlation analysis result.

[0015] In one implementation, a user terminal obtains a plurality of user interface UI atlases collected from a target application platform:

[0016] Use HTTPS to establish a connection between the client and the server;

[0017] Send data mining tasks to the user end based on the link, so that the user end can simulate human operations on the target application platform and collect the user's interface UI atlas;

[0018] A user interface UI atlas that receives user-side feedback.

[0019] In one embodiment, the method further comprises:

[0020] For each user, based on the identifier corresponding to the user and the identifier corresponding to the data mining task corresponding to the user, a unique identifier of the desensitized user behavior feature vector corresponding to the user is constructed, and the desensitized user behavior feature vector corresponding to the user and its unique identifier are associated and stored in the cloud database.

[0021] In one embodiment, the method further comprises:

[0022] Convert the desensitized user behavior feature vector corresponding to each user into a scatter plot matrix or a three-dimensional scatter plot;

[0023] The scatter plot matrix or three-dimensional scatter plot corresponding to each user is sent to a designated associated terminal, so that the scatter plot matrix or three-dimensional scatter plot corresponding to each user is displayed through the designated associated terminal.

[0024] In a second aspect, the present invention further provides a desensitized user behavior feature analysis device, which is applied to a server, and the server is connected to multiple user terminals for communication, and the device includes:

[0025] The image acquisition and preprocessing module is used to obtain the UI atlas of multiple users collected by the user end from the target application platform and preprocess the UI atlas;

[0026] A text generation module is used to generate a text data set corresponding to the preprocessed UI atlas through an image recognition model. The text data set is used to describe the content displayed by the UI atlas in text form;

[0027] A feature vector generation module is used to determine the desensitized user behavior feature vector corresponding to each user based on the text data set through a text analysis model, where the desensitized user behavior feature vector includes the user's data scores for multiple behavior feature dimensions;

[0028] The behavior analysis module is used to analyze the behavior correlation between any two users based on the desensitized user behavior feature vector corresponding to each user, and obtain the behavior correlation analysis result.

[0029] In a third aspect, the present invention further provides a server, comprising a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement any one of the methods provided in the first aspect.

[0030] In a fourth aspect, the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect.

[0031] The desensitized user behavior feature analysis method, device, server and medium provided by the present invention first obtain interface UI atlases of multiple users collected by the user end from the target application platform, and pre-process the interface UI atlases; then, through the image recognition model, a text data set corresponding to the pre-processed interface UI atlas is generated, and the text data set is used to describe the content displayed by the interface UI atlas in text form; then, through the text analysis big model, a desensitized user behavior feature vector corresponding to each user is determined based on the text data set, and the desensitized user behavior feature vector includes the data scores of the user for multiple behavior feature dimensions, and the behavior feature dimensions are determined by the natural language big model; finally, based on the desensitized user behavior feature vector corresponding to each user, the behavior correlation between any two users is analyzed to obtain the behavior correlation analysis result. During the data mining stage, the above method only needs to collect the user's interface UI atlas and use it as the basic data set for the subsequent desensitized user behavior feature analysis. Therefore, it does not involve the protection and encryption issues of the target application platform. At the same time, it can better protect user privacy during the data mining stage. In addition, the natural language big model is used to determine the behavior feature dimension and improve the quality of the behavior feature dimension. On this basis, after generating the text data corresponding to the interface UI atlas through the image recognition model, the text analysis big model is used to determine the user's data scores for multiple behavior feature dimensions based on the text data set, so as to obtain a more accurate desensitized user behavior feature vector, which helps to improve the accuracy of the user behavior feature analysis.

[0032] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A flowchart of a method for analyzing desensitized user behavior characteristics provided by an embodiment of the present invention;

[0036] Figure 2An architectural diagram of a method for analyzing desensitized user behavior characteristics provided by an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of a specific process of a desensitized user behavior feature analysis method provided by an embodiment of the present invention;

[0038] Figure 4 A schematic diagram of the structure of a desensitized user behavior feature analysis device provided by an embodiment of the present invention;

[0039] Figure 5 A schematic diagram of the structure of a server provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] The current challenges of user behavior data analysis mainly include: the division of behavioral feature dimensions, analysis of behavioral feature dimensions and scoring. The current challenges of user behavior data mining are: user privacy protection, protection and encryption of target application platform API (Application Programming Interface) and other issues. Based on this, the present invention provides a desensitized user behavior feature analysis method, device, server and medium, which can effectively protect user privacy during the data mining process, and can also effectively improve the quality of behavioral feature dimensions, thereby helping to improve the accuracy of user behavior feature analysis.

[0042] To facilitate understanding of this embodiment, a method for analyzing desensitized user behavior characteristics disclosed in an embodiment of the present invention is first described in detail. The method is applied to a server, and the server communicates with multiple user terminals. Figure 1 The flowchart of a method for analyzing desensitized user behavior characteristics is shown, and the method mainly includes the following steps S102 to S108:

[0043] Step S102, obtaining a UI atlas of multiple users collected by the user end from the target application platform, and preprocessing the UI atlas.

[0044] The UI atlas contains multiple UI images of the same user, such as the user's homepage UI (UserInterface) image, comment UI image, and follow page UI image, which are related to the interface design of the target application platform. In one example, the user terminal can simulate human operations on the target application platform to trigger the target application platform to enter the corresponding interface, and the user terminal can perform capture operations such as screenshots on the interface to obtain the user's UI image.

[0045] Furthermore, after obtaining the interface UI atlas, preprocessing can be performed on each interface UI image therein, and the preprocessing may include grayscale, binarization, denoising, smoothing, and the like.

[0046] Step S104, generating a text data set corresponding to the preprocessed interface UI atlas through an image recognition model.

[0047] Among them, the input of the image recognition model is the preprocessed interface UI image, which extracts the text part in the interface UI image, recognizes the image part in the interface UI image, and outputs a text data set. The text data set is used to describe the content displayed by the interface UI atlas in text form.

[0048] Step S106, using the text analysis big model, determine the desensitized user behavior feature vector corresponding to each user based on the text data set.

[0049] Among them, the input of the text analysis big model is a text data set, based on which the data scores of users for each behavioral feature dimension are evaluated to output a desensitized user behavior feature vector. The desensitized user behavior feature vector will include the data scores of users for multiple behavioral feature dimensions. The behavioral feature dimensions are determined by the natural language big model. The behavioral feature dimensions can be browsing content, interactive behavior, sharing behavior, etc. The behavioral feature dimensions divide the user's behavior into multiple dimensions, and the data scores indicate the user's attention level relative to each behavioral feature dimension.

[0050] Step S108, based on the desensitized user behavior feature vector corresponding to each user, the behavior correlation between any two users is analyzed to obtain a behavior correlation analysis result.

[0051] In one example, for any two users, the Pearson correlation coefficient can be used to measure the linear correlation between the desensitized user behavior feature vectors corresponding to the two users, so as to obtain the behavior correlation analysis result, which is used to indicate the degree of commonality of the behaviors between the two users. Specifically, the higher the Pearson correlation coefficient, the higher the behavioral correlation between the two users.

[0052] The desensitized user behavior feature analysis method provided by the embodiment of the present invention only needs to collect the user's interface UI atlas in the data mining stage, and use it as the basic data set for subsequent desensitized user behavior feature analysis. Therefore, it does not involve the protection and encryption issues of the target application platform. At the same time, it can better protect user privacy in the data mining stage. In addition, the natural language big model is used to determine the behavior feature dimension to improve the quality of the behavior feature dimension. On this basis, after the text data corresponding to the interface UI atlas is generated by the image recognition model, the text analysis big model is continued to be used to determine the user's data scores for multiple behavior feature dimensions based on the text data set, so as to obtain a more accurate desensitized user behavior feature vector, which helps to improve the accuracy of the user behavior feature analysis.

[0053] For ease of understanding, the embodiments of the present invention provide Figure 2 The architecture diagram of a desensitized user behavior feature analysis method shown includes a target application platform, multiple user terminals, a server terminal and a cloud server. The aforementioned desensitized user behavior feature analysis method is applied to the server terminal, and the server terminal is configured with an image recognition model, a large text analysis model and an analysis algorithm.

[0054] In one implementation, first, HTTPS is used to establish a link between the user end and the server end; then, based on the link, a data mining task is sent to the user end, so that the user end simulates human operation on the target application platform and collects the user's interface UI atlas; finally, the user's interface UI atlas fed back by the user end is received. In actual applications, a data mining task is designed, and HTTPS is used to establish a link between the user end and the server end. Each single data mining task sends a mining dimension to the user end, and the user end simulates human operation according to the mining dimension for the user end to trigger the user end to provide a corresponding interface. The user end will perform a collection action for the interface and feed back the collected interface UI atlas to the server end.

[0055] After receiving the UI atlas, the server will use the image recognition model, text analysis model and analysis algorithm to process the UI atlas to obtain the final behavior correlation analysis results. Figure 3 The specific flow chart of a method for analyzing desensitized user behavior characteristics shown in FIG. 1 includes:

[0056] First, the image recognition model is used to generate a text dataset corresponding to the preprocessed interface UI atlas.

[0057] In practical applications, the image recognition model needs to be trained in advance. Specifically: First, a large number of interface UI atlases of the target application platform are collected as basic data sets, and the interface UI atlases are grayed, binarized, denoised, and smoothed. Basic feature marking and extraction are performed on the interface UI atlases. Based on basic human cognition, user avatars, user comments, user attention, user browsing records and other dimensions are marked for image labeling. The logical dimension of this labeling is divided by the natural language large model (LLMS), and then the convolutional neural network and deep learning model are trained using the qualified data to obtain the image recognition model. The training process is as follows: forward propagation, loss function calculation, back propagation, and optimization algorithm. Forward propagation is the calculation of the input data through each layer of the neural network to finally obtain the output result. The loss function calculates the difference between the output result and the true label. Back propagation updates the weights and biases in the network according to the gradient of the loss function. The optimization algorithm uses gradient descent to adjust the network parameters during the training process. The training achieves the minimum loss to obtain an image recognition artificial intelligence model that can recognize user behavior.

[0058] Then, through the text analysis model, the desensitized user behavior feature vector corresponding to each user is determined based on the text dataset.

[0059] In practical applications, the behavioral feature dimensions need to be divided in advance. The behavioral feature dimensions can be divided based on the pre-configured user description information through the natural language large model to obtain multiple behavioral feature dimensions corresponding to the user. The user description information can be pre-configured by the staff based on the user group of the customer (such as e-commerce or social platform, etc.). In addition, the behavioral feature dimensions of different users may be different. For example, for users who are interested in kitchen renovation, the behavioral feature dimensions output by the LLMS model may include multiple dimensions such as environmental protection level and coating.

[0060] Based on the aforementioned behavioral feature dimensions, a desensitized user behavior feature vector can be generated according to the following process: For each user, a large text analysis model is used to capture the long-distance dependencies in the text data set corresponding to the user using the self-attention mechanism and position encoding, and the data scores of the user for multiple behavioral feature dimensions are determined based on the long-distance dependencies to construct the desensitized user behavior feature vector corresponding to the user. In specific implementation, the large text analysis model uses a deep learning artificial intelligence large model based on the Transformer architecture to capture long-distance dependencies in the text through self-attention mechanisms and position encoding. Let computers learn through observation and imitate the way humans acquire knowledge. Through deep neural networks, mathematical calculations are used to communicate and transmit information within the model.

[0061] The server obtains the text dataset and splits it into multi-dimensional labels using the text analysis model:

[0062] Text=Vector(V1,V2...,V n );

[0063] Among them, Text is long text data, Vector is a multi-dimensional vector, and V is the dimension score (0-100). The above formula can be used to obtain the data score of the user's own behavior characteristics.

[0064] Furthermore, the server is also connected to the cloud database. On this basis, for each user, based on the identifier corresponding to the user and the identifier corresponding to the data mining task corresponding to the user, a unique identifier of the desensitized user behavior feature vector corresponding to the user is constructed, and the desensitized user behavior feature vector corresponding to the user and its unique identifier are associated and stored in the cloud database. Specifically, a cloud database is established, a multi-user dimensional vector storage structure is designed, and a scoring data storage structure is established, in which the user's ID and the data mining task ID constitute the unique ID of the desensitized user behavior feature vector, and the desensitized user behavior feature vector and its unique ID are stored in the cloud database.

[0065] Finally, based on the desensitized user behavior feature vector corresponding to each user, the behavior correlation between any two users is analyzed to obtain the behavior correlation analysis results.

[0066] In one embodiment, for any two users, the Pearson correlation coefficient between the first user and the second user is determined based on the data scores corresponding to each behavioral feature dimension in the desensitized user behavioral feature vectors corresponding to the first user and the second user, and the Pearson correlation coefficient is used as the behavioral correlation analysis result. Specifically, the behavioral correlation analysis result between the first user and the second user can be determined according to the following formula:

[0067]

[0068] Among them, R is the Pearson correlation coefficient, whose value is between -1 and 1, and x i is the data score corresponding to the i-th behavior feature dimension in the desensitized user behavior feature vector corresponding to the first user, is the mean of the data scores corresponding to each behavioral feature dimension in the desensitized user behavior feature vector corresponding to the first user, y i is the data score corresponding to the i-th behavior feature dimension in the desensitized user behavior feature vector corresponding to the second user, It is the mean of the data scores corresponding to each behavior feature dimension in the desensitized user behavior feature vector corresponding to the second user.

[0069] Furthermore, the embodiment of the present invention can also convert the desensitized user behavior feature vector corresponding to each user into a scatter plot matrix or a three-dimensional scatter plot, and send it to a designated associated terminal, so as to display the scatter plot matrix or the three-dimensional scatter plot corresponding to each user through the designated associated terminal. In practical applications, users within the range of the same dimensional label can be extracted for numerical analysis, and the relationship between each pair of feature vectors can be displayed through the scatter plot matrix and the three-dimensional scatter plot analysis of the user characteristics of multidimensional data.

[0070] In summary, the construction system of the embodiment of the present invention includes a user end and a remote server artificial intelligence analysis system (that is, the aforementioned server end). The user end is used to mine the behavioral characteristics of desensitized users. The remote server artificial intelligence system receives the instructions and data information of the data user end and is responsible for analyzing the user's desensitized behavior data for effective analysis and scoring. The remote server platform and the data user end use an HTTPS link for collaborative data exchange. The cloud database of the remote server is responsible for storing the desensitized user behavior feature vector after analysis, which is used to analyze the user's interests, behaviors, preferences, etc. The desensitized user behavior feature analysis method of the embodiment of the present invention relies on the comprehensive analysis of artificial intelligence algorithms and feature vectors, and can efficiently and quickly analyze the characteristics of users, help enterprises more accurately locate target customers and analyze user preferences and relationships, and improve the quality of data analysis data.

[0071] The embodiments of the present invention have at least the following characteristics: (1) During the data mining process, the embodiments of the present invention will not involve issues such as protection and encryption of platform APIs; (2) During the data mining process, the embodiments of the present invention effectively protect user privacy; (3) The determination and division of user behavior feature dimensions in the embodiments of the present invention are automatically generated by AI (Artificial Intelligence) with higher dimensional quality, i.e., the aforementioned large natural language model (LLMS); (4) The analysis and scoring of the behavior feature dimensions in the embodiments of the present invention are based on the AI ​​knowledge base with higher quality and accuracy.

[0072] Based on the above embodiments, the present invention provides a desensitized user behavior feature analysis device, which is applied to a server, and the server is connected to multiple user terminals for communication. Figure 4 The structure diagram of a desensitized user behavior feature analysis device shown in FIG. 1 mainly includes the following parts:

[0073] The image acquisition and preprocessing module 402 is used to obtain the UI atlas of multiple users collected by the user end from the target application platform and preprocess the UI atlas;

[0074] The text generation module 404 is used to generate a text data set corresponding to the preprocessed UI atlas through an image recognition model, and the text data set is used to describe the content displayed by the UI atlas in text form;

[0075] A feature vector generating module 406 is used to determine a desensitized user behavior feature vector corresponding to each user based on a text data set by using a text analysis large model, wherein the desensitized user behavior feature vector includes data scores of the user for multiple behavior feature dimensions;

[0076] The behavior analysis module 408 is used to analyze the behavior correlation between any two users based on the desensitized user behavior feature vector corresponding to each user, and obtain a behavior correlation analysis result.

[0077] The desensitized user behavior feature analysis device provided by the embodiment of the present invention only needs to collect the user's interface UI atlas in the data mining stage, and use it as the basic data set for subsequent desensitized user behavior feature analysis. Therefore, it does not involve the protection and encryption issues of the target application platform. At the same time, it can better protect user privacy in the data mining stage. In addition, the natural language big model is used to determine the behavior feature dimension to improve the quality of the behavior feature dimension. On this basis, after the text data corresponding to the interface UI atlas is generated by the image recognition model, the text analysis big model is continued to be used to determine the user's data scores for multiple behavior feature dimensions based on the text data set, so as to obtain a more accurate desensitized user behavior feature vector, which helps to improve the accuracy of the user behavior feature analysis.

[0078] In one embodiment, a dimension division module is further included, which is used to:

[0079] Through the natural language big model, the behavioral feature dimensions are divided based on the pre-configured user description information to obtain multiple behavioral feature dimensions corresponding to the user.

[0080] In one implementation, the feature vector generation module 406 is specifically used to:

[0081] For each user, a large text analysis model is used to capture the long-distance dependencies in the text dataset corresponding to the user using the self-attention mechanism and position encoding. Based on the long-distance dependencies, the data scores of the user for multiple behavioral feature dimensions are determined to construct a desensitized user behavior feature vector corresponding to the user.

[0082] In one implementation, the behavior analysis module 408 is specifically used to:

[0083] For any two users, the Pearson correlation coefficient between the first user and the second user is determined based on the data scores corresponding to each behavior feature dimension in the desensitized user behavior feature vectors corresponding to the first user and the second user respectively, and the Pearson correlation coefficient is used as the behavior correlation analysis result.

[0084] In one implementation, the image acquisition and preprocessing module 402 is specifically used to:

[0085] Use HTTPS to establish a connection between the client and the server;

[0086] Send data mining tasks to the user end based on the link, so that the user end can simulate human operations on the target application platform and collect the user's interface UI atlas;

[0087] A user interface UI atlas that receives user-side feedback.

[0088] In one embodiment, the device further includes a storage module for:

[0089] For each user, based on the identifier corresponding to the user and the identifier corresponding to the data mining task corresponding to the user, a unique identifier of the desensitized user behavior feature vector corresponding to the user is constructed, and the desensitized user behavior feature vector corresponding to the user and its unique identifier are associated and stored in the cloud database.

[0090] In one embodiment, a display module is further included for:

[0091] Convert the desensitized user behavior feature vector corresponding to each user into a scatter plot matrix or a three-dimensional scatter plot;

[0092] The scatter plot matrix or three-dimensional scatter plot corresponding to each user is sent to a designated associated terminal, so that the scatter plot matrix or three-dimensional scatter plot corresponding to each user is displayed through the designated associated terminal.

[0093] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0094] An embodiment of the present invention provides a server, specifically, the server includes a processor and a storage device; the storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0095] Figure 5A structural diagram of a server provided in an embodiment of the present invention, the server 100 includes: a processor 50, a memory 51, a bus 52 and a communication interface 53, wherein the processor 50, the communication interface 53 and the memory 51 are connected via the bus 52; the processor 50 is used to execute an executable module stored in the memory 51, such as a computer program.

[0096] The memory 51 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 53 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0097] The bus 52 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0098] Among them, the memory 51 is used to store programs, and the processor 50 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 50 or implemented by the processor 50.

[0099] The processor 50 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 50. The above processor 50 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to be executed, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 51, and the processor 50 reads the information in the memory 51 and completes the steps of the above method in combination with its hardware.

[0100] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0101] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0102] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for analyzing desensitized user behavior characteristics, characterized in that: The method is applied to a server, the server is connected to multiple clients for communication, and the method includes: Obtaining a UI atlas of multiple users collected by the user end from the target application platform, and preprocessing the UI atlas; Generate a text data set corresponding to the preprocessed UI atlas through an image recognition model, wherein the text data set is used to describe the content displayed by the UI atlas in text form; Determine, by means of a text analysis big model, a desensitized user behavior feature vector corresponding to each of the users based on the text data set, wherein the desensitized user behavior feature vector includes data scores of the user for multiple behavior feature dimensions, and the behavior feature dimensions are determined by means of a natural language big model; Based on the desensitized user behavior feature vector corresponding to each of the users, the behavior correlation between any two of the users is analyzed to obtain a behavior correlation analysis result.

2. The method for analyzing desensitized user behavior characteristics according to claim 1, characterized in that: Before determining the desensitized user behavior feature vector corresponding to each user based on the text data set through the text analysis big model, the method further includes: Through the natural language big model, the behavioral feature dimensions are divided based on the pre-configured user description information to obtain multiple behavioral feature dimensions corresponding to the user.

3. The method for analyzing desensitized user behavior characteristics according to claim 1, characterized in that: The text analysis model adopts a deep learning model based on the Transform architecture; Determining the desensitized user behavior feature vector corresponding to each user based on the text data set through a text analysis model includes: For each of the users, a large text analysis model is used to capture the long-distance dependencies in the text data set corresponding to the user using the self-attention mechanism and position encoding. Based on the long-distance dependencies, the data scores of the user for multiple behavioral feature dimensions are determined to construct a desensitized user behavior feature vector corresponding to the user.

4. The method for analyzing desensitized user behavior characteristics according to claim 1, characterized in that: Based on the desensitized user behavior feature vector corresponding to each of the users, the behavior correlation between any two of the users is analyzed to obtain a behavior correlation analysis result, including: For any two of the users, based on the data scores corresponding to each of the behavioral feature dimensions in the desensitized user behavior feature vectors corresponding to the first user and the second user respectively, the Pearson correlation coefficient between the first user and the second user is determined, and the Pearson correlation coefficient is used as the behavioral correlation analysis result.

5. The method for analyzing desensitized user behavior characteristics according to claim 1, characterized in that: Get the UI atlas of multiple users collected by the user end from the target application platform: Using HTTPS to establish a link between the client and the server; Sending a data mining task to the user terminal based on the link, so that the user terminal simulates human operation on the target application platform and collects the user's interface UI atlas; Receive the user's interface UI atlas fed back by the user terminal.

6. The method for analyzing desensitized user behavior characteristics according to claim 5, characterized in that: The method further comprises: For each of the users, based on the identifier corresponding to the user and the identifier corresponding to the data mining task corresponding to the user, a unique identifier of the desensitized user behavior feature vector corresponding to the user is constructed, and the desensitized user behavior feature vector corresponding to the user and its unique identifier are associated and stored in the cloud database.

7. The method for analyzing desensitized user behavior characteristics according to claim 1, characterized in that: The method further comprises: Converting the desensitized user behavior feature vector corresponding to each of the users into a scatter plot matrix or a three-dimensional scatter plot; The scatter plot matrix or the three-dimensional scatter plot corresponding to each of the users is sent to a designated associated terminal, so that the scatter plot matrix or the three-dimensional scatter plot corresponding to each of the users is displayed through the designated associated terminal.

8. A desensitized user behavior feature analysis device, characterized in that: The device is applied to a server, the server is connected to multiple clients for communication, and the device includes: An image acquisition and preprocessing module, used to obtain a plurality of user interface UI atlases collected by the user end from the target application platform, and preprocess the interface UI atlases; A text generation module, used to generate a text data set corresponding to the preprocessed interface UI atlas through an image recognition model, wherein the text data set is used to describe the content displayed by the interface UI atlas in text form; A feature vector generation module, used to determine a desensitized user behavior feature vector corresponding to each of the users based on the text data set through a text analysis model, wherein the desensitized user behavior feature vector includes data scores of the user for multiple behavior feature dimensions; The behavior analysis module is used to analyze the behavior correlation between any two of the users based on the desensitized user behavior feature vector corresponding to each of the users, and obtain a behavior correlation analysis result.

9. A server, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 7.

Images