A user analysis method and related device

Abstract

An embodiment of the present application discloses a user analysis method and related devices, which relate to the technical field of data analysis. The method includes: obtaining user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user; obtaining an interaction object vector corresponding to the target interaction object; and determining a security parameter corresponding to the target user according to the interaction object vector, where the security parameter is used to identify the probability of the target user having a risk interaction behavior. While enriching the types of analysis data, this method introduces the interaction order of the interaction object as a new analysis criterion, further improving the accuracy and rationality of the analysis.

Description

Technical Field

[0001] This application relates to the technical field of data analysis, and in particular, to a user analysis method and related devices. Background Art

[0002] Predicting possible behaviors of users based on their interaction behavior data is a common user analysis method. In related technologies, numerical vectors with a certain time sequence, such as user transaction time and transaction amount, are usually obtained, and users are analyzed through a time series model. However, for some other types of information that do not contain time sequence information, the time series model cannot understand well, resulting in a relatively single type of data and low analysis accuracy. Summary of the Invention

[0003] To solve the above technical problems, an embodiment of this application provides a user analysis method. The processing device can generate a time-sequential interaction object vector for each interaction object based on the interaction order of the user among interaction objects. This interaction object vector can be used for risk behavior analysis of the user, thereby enriching the types of analysis data while introducing the interaction order of the interaction objects as a new analysis criterion, further improving the accuracy and rationality of the analysis.

[0004] The embodiments of this application disclose the following technical solutions:

[0005] In a first aspect, an embodiment of this application discloses a user analysis method, and the method includes:

[0006] Obtain user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user;

[0007] Obtain an interaction object vector corresponding to the target interaction object, where the interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users among multiple interaction objects, and the multiple interaction objects include the target interaction object;

[0008] Determine a security parameter corresponding to the target user according to the interaction object vector, where the security parameter is used to identify the probability that the target user has a risk interaction behavior.

[0009] In a second aspect, an embodiment of this application discloses a user analysis device, and the method includes a first acquisition unit, a second acquisition unit, and a determination unit:

[0010] The first acquisition unit is configured to obtain user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user;

[0011] The second acquisition unit is configured to acquire an interaction object vector corresponding to the target interaction object. The interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users with multiple interaction objects. The multiple interaction objects include the target interaction object.

[0012] The determination unit is configured to determine a security parameter corresponding to the target user according to the interaction object vector. The security parameter is used to identify the probability of the target user having a risky interaction behavior.

[0013] In a third aspect, an embodiment of the present application provides a computer device, which includes a processor and a memory:

[0014] The memory is configured to store program code and transmit the program code to the processor;

[0015] The processor is configured to execute the user analysis method described in the first aspect according to the instructions in the program code.

[0016] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which is configured to store a computer program, and the computer program is used to execute the user analysis method described in the first aspect.

[0017] It can be seen from the above technical solutions that, in order to incorporate the information of the interaction object, which does not have temporal order itself, into the analysis of the user, an interaction object vector with temporal order can be generated based on the interaction order of the user with multiple interaction objects reflected in the historical user behavior data. After acquiring the user behavior data corresponding to the target user, these interaction object vectors can be used to represent the target interaction object identified in the user behavior data, and the interaction object information that originally does not have temporal order can be transformed into an interaction object vector with temporal order, so that the target user can be analyzed based on more dimensional data, improving the accuracy of user analysis. At the same time, since the historical user behavior data can reflect the interaction order of multiple users with respect to the interaction object, the interaction object vector obtained based on the historical user behavior data can reflect the interaction objects that the user may interact with in the time periods before and after interacting with the interaction object under normal circumstances. Therefore, based on the interaction object vector corresponding to the target interaction object, it can be analyzed whether the user behavior of the target user conforms to the user behavior under normal circumstances, and the security parameter corresponding to the target user can be determined. The security parameter is used to identify the probability of the target user having a risky interaction behavior. Description of the Drawings

[0018] 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.

[0019] Figure 1 Schematic diagram of a user analysis method in an actual application scenario provided by an embodiment of the present application;

[0020] Figure 2 Flowchart of a user analysis method provided by an embodiment of the present application;

[0021] Figure 3 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0022] Figure 4 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0023] Figure 5 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0024] Figure 6 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0025] Figure 7 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0026] Figure 8 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0027] Figure 9 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0028] Figure 10 Schematic diagram of a user analysis method provided by an embodiment of the present application;

[0029] Figure 11 Schematic diagram of a user analysis method in an actual application scenario provided by an embodiment of the present application;

[0030] Figure 12 Schematic diagram of a user analysis method in an actual application scenario provided by an embodiment of the present application;

[0031] Figure 13 Block diagram of the structure of a user analysis device provided by an embodiment of the present application;

[0032] Figure 14The structural diagram of a computer device provided by an embodiment of the present application;

[0033] Figure 15 The structural diagram of a server provided by an embodiment of the present application. Specific embodiments

[0034] The embodiments of the present application will be described below with reference to the accompanying drawings.

[0035] How to analyze users more accurately is one of the key issues concerned by relevant personnel. In the related art, the way to analyze whether a user will have a risk interaction behavior is mainly based on the user's interaction time and interaction amount. These information are all numerical information and have a certain time series. Therefore, the time series model can analyze users more accurately based on this information.

[0036] However, information of text types such as merchant names and product names usually does not have a time series. If such information is directly input into the time series model, it may cause the model to be difficult to understand the content of this information, and thus unable to analyze users accurately.

[0037] To solve the above technical problems, an embodiment of the present application provides a user analysis method. The processing device can generate a time-series interaction object vector for each interaction object based on the interaction order between interaction objects. This interaction object vector can be used to analyze the risk behavior of users, thereby enriching the types of analysis data and introducing the interaction order of interaction objects as a new analysis criterion, further improving the accuracy and rationality of the analysis.

[0038] It can be understood that this method can be applied to a processing device, which is a processing device with data analysis functions. For example, it can be a terminal device or a server with data analysis functions. This method can be independently executed by a terminal device or a server, or can also be applied to a network scenario where a terminal device and a server communicate, and run in cooperation with the terminal device and the server. Among them, the terminal device can be a mobile phone, a desktop computer, a personal digital assistant (Personal Digital Assistant, abbreviated as PDA), a tablet computer and other devices. The server can be understood as an application server or a Web server. In actual deployment, the server can be an independent physical server or a server cluster or distributed system composed of multiple physical servers. The terminal and the server can be directly or indirectly connected through wired or wireless communication methods, and the present application does not limit this here.

[0039] The embodiments of the present application can also be applied with blockchain technology. For example, in the user analysis method disclosed in the present application, multiple servers can be used to determine interaction object vectors and other processes. Among them, multiple servers can form a blockchain, and the server is a node on the blockchain.

[0040] In addition, the present application also relates to Artificial Intelligence (AI) technology. Artificial Intelligence is to use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results in theory, methods, technologies, and application systems. 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 react 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.

[0041] Artificial Intelligence technology is an interdisciplinary subject, involving a wide range of fields, including both hardware-level technologies and software-level technologies. Artificial Intelligence basic technologies generally include technologies such as sensors, dedicated Artificial Intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial Intelligence software technologies mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning. Among them, the present application mainly relates to natural language processing technology and machine learning technology.

[0042] Natural Language Processing (NLP) is an important direction in the field of computer science and Artificial Intelligence. It studies various theories and methods that can achieve effective communication between humans and computers in natural language. Natural Language Processing is a science that integrates linguistics, computer science, and mathematics. Therefore, the research in this field will involve natural language, that is, the language people use in daily life, so it has a close connection with the research of linguistics. Natural Language Processing technology usually includes technologies such as text processing, semantic understanding, machine translation, robot question answering, and knowledge graphs.

[0043] Machine Learning (ML) is an interdisciplinary field that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and 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] In the embodiments of the present application, the processing device can process the relevant information of the interaction object through natural language processing technology, accurately determine the interaction object vector through machine learning technology, and at the same time, a time series model for determining the user security parameter can be trained.

[0045] To facilitate the understanding of the technical solution of the present application, next, a user analysis method provided in the embodiments of the present application will be introduced in combination with an actual application scenario.

[0046] See Figure 1 , Figure 1 which is a schematic diagram of a user analysis method in an actual application scenario provided in the embodiments of the present application. In this actual application scenario, the processing device is server 101, and the server 101 can analyze whether the user will have a risk interaction behavior based on the information related to the user.

[0047] When performing user analysis on the target user, the server 101 can first obtain the user behavior data corresponding to the target user, and the user behavior data can be used to identify the target interaction object associated with the target user. In order to be able to accurately express the target interaction object by using time-sequential information and make the target interaction object a kind of information for analyzing the target user, the server 101 can obtain historical behavior data, which can reflect the interaction order of multiple users such as User A, User B, and User C among multiple interaction objects such as Interaction Object A, Interaction Object B, Interaction Object C, and the target interaction object. Since the interaction order belongs to time-sequential information, the server 101 can generate an interaction object vector corresponding to the target interaction object based on the interaction order, and the interaction object vector can be used to analyze the target user.

[0048] Meanwhile, since the historical user behavior data can reflect the interaction order of multiple users with multiple interaction objects, the interaction object vector generated based on the historical user behavior data can reflect the interaction objects that most users may interact with before and after when interacting with the target interaction object. Furthermore, the server 101 can analyze whether a risk interaction behavior may occur when the target user conducts a series of interactions with multiple interaction objects including the target interaction object based on this information, so as to determine the security parameter corresponding to the target user, and this security parameter is used to identify the probability of the target user having a risk interaction behavior.

[0049] It can be seen that through the above technical solution, the server 101 can more accurately express the interaction object by using the interaction object vector with time series, so as to analyze the user based on more dimensional information. At the same time, the interaction object vector can reflect the interaction order characteristics of users in the interaction object under normal circumstances. Furthermore, it can analyze whether the interaction behavior of the target user in the interaction object conforms to the normal behavior characteristics of users, and more accurately determine the security parameter corresponding to the target user.

[0050] Next, a user analysis method provided by an embodiment of the present application will be introduced in conjunction with the accompanying drawings.

[0051] See Figure 2 , Figure 2 which is a flowchart of a user analysis method provided by an embodiment of the present application. The method includes:

[0052] S201: Obtain the user behavior data corresponding to the target user.

[0053] Among them, the target user can be any user who has had an interaction behavior, and the user behavior data is used to identify the target interaction object associated with the target user. It can be understood that for different user interaction scenarios, the interaction object can also include various objects. For example, when analyzing the transaction behavior of a user in an entity transaction, the interaction object can be the merchant that the user has consumed; when analyzing the transaction behavior of a user in an online shopping transaction, the interaction object can be the goods that the user has browsed or purchased, etc.

[0054] S202: Obtain the interaction object vector corresponding to the target interaction object.

[0055] Since the interaction object is usually text information without time series such as merchant names and product names, if the interaction object is directly input into the time series model for analysis, it will cause the model to be unable to accurately understand the meaning of this information, and thus unable to accurately analyze the user.

[0056] Based on this, in order to incorporate interaction object information into the analysis of users by the time series model, the processing device can obtain temporal information related to the interaction object and generate input information for characterizing the interaction object based on this temporal information. It can be understood that when a user interacts with multiple interaction objects, there is usually a certain interaction order, and the interaction order itself is a kind of temporal information. Therefore, in the embodiments of this application, the processing device can generate the input information based on the interaction order of the user among multiple interaction objects. Among them, multiple interaction objects refer to two or more interaction objects.

[0057] First, the processing device can obtain historical user behavior data, which is used to reflect the interaction order of multiple users among multiple interaction objects, and the multiple interaction objects include the target interaction object. Thus, based on the historical user behavior data, the processing device can determine an interaction object vector for each interaction object among the multiple interaction objects, and the interaction object vector is used to reflect the characteristics of the corresponding interaction object in terms of the interaction order among the multiple interaction objects. After obtaining the user behavior data, the processing device can determine the target interaction object identified by the user behavior data, and then obtain the interaction object vector corresponding to the target interaction object, and the interaction object vector can characterize the target interaction object from the temporal level.

[0058] S203: Determine the security parameter corresponding to the target user according to the interaction object vector.

[0059] As mentioned above, the historical user behavior data can be used to reflect the interaction order of multiple users among multiple interaction objects. Therefore, based on the historical behavior data, the processing device can, to a certain extent, know the interaction order habits of users among multiple interaction objects under normal circumstances. Thus, the interaction object vector determined based on the historical user behavior data can, to a certain extent, reflect other interaction objects that the target user may interact with in the time periods before and after interacting with the target interaction object under normal circumstances. Furthermore, based on the interaction object vector, the processing device can analyze the user behavior data of the target user, that is, analyze whether the interaction behavior of the target user among multiple interaction objects conforms to the user behavior under normal circumstances.

[0060] The processing device can determine the security parameter corresponding to the target user based on this user analysis. The security parameter is used to identify the probability of the target user having a risky interaction behavior, and the risky interaction behavior refers to an interaction behavior with a certain security risk, such as loan overdue, risky payment behavior, etc. It can be understood that the closer the interaction behavior of the target user is to the interaction behavior of ordinary users, the lower the probability of the target user having a risky interaction behavior, and vice versa.

[0061] As can be seen from the above technical solution, in order to incorporate information about interaction objects, which inherently has no temporal sequence, into the analysis of users, interaction object vectors with temporal sequence can be generated based on the interaction order of users among multiple interaction objects reflected in historical user behavior data. After obtaining the user behavior data corresponding to the target user, these interaction object vectors can be used to represent the target interaction objects identified in the user behavior data, converting the interaction object information that originally has no temporal sequence into interaction object vectors with temporal sequence, so that user analysis of the target user can be performed based on data from more dimensions, improving the accuracy of user analysis. At the same time, since the historical user behavior data can reflect the interaction order of multiple users with respect to interaction objects, the interaction object vectors obtained based on this historical user behavior data can reflect the interaction objects that users are likely to interact with in the periods before and after interacting with this interaction object under normal circumstances. Therefore, based on the interaction object vector corresponding to the target interaction object, it can be analyzed whether the user behavior of the target user conforms to the user behavior under normal circumstances, and the security parameter corresponding to the target user can be determined, and this security parameter is used to identify the probability of the target user having a risky interaction behavior.

[0062] In practical applications, there are various ways for the processing device to obtain the interaction object vector corresponding to the target interaction object based on historical user behavior data. In one possible implementation, to improve the determination accuracy of the interaction object vector, the processing device can introduce the Node2vec algorithm and determine the interaction object vector through random walk.

[0063] First, the processing device can generate an interaction object network based on the historical user behavior data. The interaction object network includes interaction object nodes corresponding to multiple interaction objects respectively, and directed connections for connecting the interaction object nodes. The directed connections are generated based on the interaction order of multiple users among multiple interaction objects. That is, if there is a user who first interacts with interaction object A and then with interaction object B, a directed connection pointing from the interaction object A node to the interaction object B node can be generated in the interaction object network.

[0064] As Figure 3 shown, first, through the historical user behavior data, the processing device can determine the interaction order of multiple users among interaction objects A, B, C, D, E, and F, and then an interaction object network as shown in the figure can be formed based on this interaction order. The interaction object network includes six nodes corresponding to six interaction objects and directed connections between the nodes. If a certain user first interacts with interaction object A and then with interaction object B, the processing device can connect a line between the A node and the B node.

[0065] Meanwhile, to further improve the accuracy of user analysis, the directed connection can also be used to reflect the interaction weights between the connected interactive object nodes. The interaction weight represents the frequency of interactions that occur between the user and the interactive object corresponding to the connected interactive object node. For example, in the historical user behavior data, if one user has an interaction behavior with interactive object A first and then with interactive object B, the weight of the directed connection from node A to node B is 1. Similarly, if the above interaction behavior occurs again, the weight of the directed connection is incremented by 1. Thus, through the weight of the directed connection, the processing device can also learn the interaction frequency information among multiple interactive objects of the user, further enriching the information content and improving the accuracy of user analysis. The processing device can determine the interactive object vectors corresponding to multiple interactive objects based on the interactive object network, and the multiple interactive objects include the target interactive object. Thus, the processing device can obtain the interactive object vector corresponding to the target interactive object.

[0066] Specifically, in a possible implementation manner, after determining the interactive object network, the processing device can generate the interactive object vectors of the corresponding nodes through a Random Walk strategy. To execute the Random Walk strategy, the processing device can first set the walk parameters for executing the Random Walk strategy. In Random Walk, there are two strategies, one is Breadth First Search (BFS) and the other is Depth First Search (DFS). As Figure 4 shown, it can be considered that node U is similar to nodes S1, S2, S3, and S4. They are the direct neighbors in the network and can be called Homophily. It can also be considered that node U is more similar to node S6, which is called structural equivalence. To find the direct neighbors starting from a node, BFS can be used. If you want to find structurally similar ones, you need to "go out", so DFS needs to be used.

[0067] Among them, the walk parameters are used to control the walk tendency of the Random Walk strategy, that is, whether it is more inclined to BFS or DFS. As Figure 5 shown, assume that we start a random walk from node t and now reach node v. If the BFS strategy is adopted, we should go to x1 because both v and x1 are the direct neighbors of node t. If the DFS strategy is adopted, we should go to x2 or x3 because they are both one step away from t. Of course, it is also possible to return to node t. Thus, the transition probability π vx is:

[0068] π vx = α pq(t, x) · ω vx

[0069] where ω vx is the weight of the connection between two nodes, i.e., the edge weight in the network. α pq (t, x) is the probability search bias for selecting each walking strategy, defined as:

[0070]

[0071] How a node moves to the next step depends on the relationship between its previous step and the next step.

[0072] v is the current node, t is the node where the previous step of v is located, and x represents the position of the next step. d tx represents the shortest distance between t and x:

[0073] When d tx = 0, it means returning from v back to node t. At this time, the search bias is 1 / p, which can be understood as returning to the previous step with a probability of 1 / p;

[0074] When d tx = 1, then x is a direct neighbor of t, equivalent to BFS. At this time, the search bias is 1;

[0075] When d tx = 2, then x is a neighbor of a neighbor of t, equivalent to DFS. At this time, the search bias is 1 / q;

[0076] p is the return parameter, because p controls the probability of returning to the original node; q is the in - out parameter, because it controls the relationship between BFS and DFS.

[0077] After setting the walking parameters, the processing device can generate a set of interaction object sequences according to the interaction object network by setting the random walking strategy of the walking parameters. The set of interaction object sequences includes multiple interaction object sequences, and the interaction object sequences are used to represent the possible interaction orders when the user interacts among the interaction objects included in the interaction object network.

[0078] For example, the processing device can set different p and q to obtain different - weighted interaction object sequences. When training the model, the grid search method can be used to find the optimal p and q, or p and q can be selected according to the needs of the scenario. For example, in a possible implementation, the BFS strategy can be mainly adopted, where p is 1 and q is 2, for random walking to generate an interaction object sequence with a length of 10.

[0079] Based on the set of interaction object sequences, the processing device can determine the interaction object vectors corresponding to multiple interaction objects through a word vector model, so that the characteristics of the dimension of the interaction order can be more fully incorporated into the interaction object vectors. For example, the word vector model can be a Word2vec model. The processing device can use the set of interaction object sequences as input and input it into the Word2vec model. By adopting the skip-gram of the Word2vec model, the context is predicted through the central word, that is, the previous and subsequent interaction objects in an interaction object sequence are predicted, and the final interaction object vector is output through Word2vec.

[0080] In addition to the information of this type of interaction object, in order to further improve the accuracy of user analysis, the processing device can also incorporate more types of data information. In a possible implementation manner, the user behavior data can be the user behavior data corresponding to the target user within the target time interval, and the target time interval can be any time interval.

[0081] In addition to obtaining the user behavior data, the processing device can also obtain the interaction data corresponding to the target user within the target time interval, and the interaction data is used to identify any combination of the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user.

[0082] When determining the security parameter, the processing device can determine the security parameter corresponding to the target user according to the interaction object vector and the interaction data, so that more types of data such as the time interval and interaction amount between multiple interaction behaviors can be incorporated into the analysis of the user, and the accuracy of user analysis is further improved.

[0083] In order to enable the processing device to perform a more reasonable analysis of the above data, in a possible implementation manner, if the interaction data is used to identify the time interval between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user, the processing device can determine the interaction time interval and interaction amount corresponding to the target object according to the interaction data. The interaction time interval is used to identify the time interval for the target user to interact between adjacent interaction objects in time sequence, and the interaction amount is the interaction amount corresponding to the target interaction object. Thus, the processing device can first obtain the relevant information corresponding to the target interaction object from the interaction data.

[0084] Subsequently, the processing device can normalize the interaction time interval and the interaction amount, enabling the relevant model to more clearly identify the data. The processing device can splice the normalized interaction time interval and the normalized interaction amount with the interaction object vector corresponding to the target interaction object to generate a fused interaction vector, which can more comprehensively reflect the information related to the target interaction object. The processing device can determine the security parameter corresponding to the target user based on the fused interaction vector, so that on the basis of enriching the types of information, it can classify and splice the information targeted at the interaction object, and then can analyze the user more accurately.

[0085] Of course, when the interaction data is only used to identify the time interval between multiple interaction behaviors corresponding to the target user, the processing device can only determine the interaction time interval corresponding to the target object according to the interaction data, and generate a fused interaction vector based on the interaction time interval; when the interaction data is only used to identify the interaction amount of multiple interaction behaviors corresponding to the target user, the processing device can only determine the interaction amount corresponding to the target object according to the interaction data, and generate a fused interaction vector based on the interaction amount.

[0086] Among them, the way for the processing device to determine the security parameter according to the fused interaction vector can also include various methods. In one possible implementation, in order to conform to the characteristics of data information timeliness, the processing device can determine the security parameter corresponding to the target user through a time series model based on the fused interaction vector.

[0087] For example, the time series model can be a Long Short-Term Memory (LSTM) network. The processing device can use the fused interaction vector as the input of the LSTM model. As Figure 6 shown, among them, the interaction object vector is the merchant embedding vector, the interaction time interval is the time interval between two transactions, and the interaction amount is the amount in the figure. By splicing multiple interaction object vectors with their corresponding interaction time intervals and interaction amounts, a sequence of fused interaction vectors can be obtained, that is, the user transaction sequence shown in the figure, and this user transaction sequence can be used as the input Input of the LSTM model.

[0088] Among them, the LSTM model includes three gates, namely the forget gate, the input gate, and the output gate.

[0089] (1) Forget gate

[0090] The forget gate determines what information needs to be discarded. As Figure 7 shown, this gate will read h t-1 and x t In this solution, x tThat is, the user transaction sequence [v1, v2, v3…v 33 , v 34 , where σ is the sigmoid function. After passing through σ, a value between 0 and 1 is obtained for each number in C t-1 . The specific formula is as follows:

[0091] f t = σ(W f ·[h t-1 , x t + b f )

[0092] Among them, b f and W f are the parameters of σ, h t-1 is the output result of the output gate in the previous iteration process, f t is the number assigned by the forget gate to C t-1 , and C t-1 is the cell state updated by the input gate in the previous iteration process. During the training process of the LSTM model, these b f and W f parameters can be adjusted accordingly.

[0093] (2) Input gate

[0094] The input gate determines how much new information is added to the new state. As Figure 8 shown, first, h t-1 and x t pass through a sigmoid function to determine which information needs to be updated; a tanh layer generates a vector, which is the alternative content for update In the next step, multiply these two parts to update the cell state. As shown in the following formula, where b i and W i are the relevant parameters of the function σ, and W C and b c are the relevant parameters of the function tanh:

[0095] i t = σ(W i ·[h t-1 , x t + b i )

[0096]

[0097] Subsequently, as Figure 9 shown, multiply C t-1 by f t obtained through the forget gate, discard the unnecessary information, and then add A new candidate value is obtained, and the formula is as follows:

[0098]

[0099] Among them, C t is the cell state after the input gate is updated in this iteration. The input gate can pass this updated state to the output gate so that the output gate can calculate the output value h t of this iteration.

[0100] (3) Output gate

[0101] Finally, we need to determine what value to output. As Figure 10 shown, first, h t-1 and x t pass through a sigmoid function to determine which information needs to be output. Then, C t is processed through tanh and multiplied by the output of the sigmoid gate to output the part that we determine to output. In this solution, after extracting h t as the output and inputting it into the LR model to obtain a probability value output, this probability value is the security parameter corresponding to this user. The relevant formula is as follows:

[0102] o t = σ(W o [h t-1 , x t +b o )

[0103] h t = o t *tanh(C t )

[0104] Among them, o t is the part of the information that needs to be output in h t-1 and x t . b o and W o are the relevant parameters of this σ function, and h t is the parameter output by the output gate in this iteration.

[0105] In order to introduce the technical solution provided by this application more vividly, next, a user analysis method provided by an embodiment of this application will be introduced in combination with an actual application scenario. See Figure 11 , Figure 11 which is a schematic diagram of a user analysis method in an actual application scenario provided by an embodiment of this application. In this actual application scenario, the processing device is a user analysis server.

[0106] First, the user analysis server can obtain the transaction records of a certain user. Through these transaction records, the transaction merchants, transaction amounts, and time intervals between adjacent transactions of this user can be determined. The user analysis server can perform normalization processing on the transaction time intervals and transaction amounts, and then input the merchant information into the Node2vec model as shown in Figure 12 to generate corresponding merchant vectors.

[0107] The user analysis server can splice the normalized transaction amounts and time intervals with the merchant vectors to generate a fused interaction vector, and one fused interaction vector is used to represent a single transaction. The processing device can generate a user transaction sequence based on multiple fused interaction vectors and input it into the LSTM model to obtain the final security parameter Score. This security parameter can be used to evaluate the possibility of the user defaulting in the future based on the user's loan or credit card transaction data after the user's loan application or credit card application is approved. In addition, this method can also be used in other scenarios, which are not limited here.

[0108] Based on the user analysis method provided in the above embodiments, the embodiments of the present application also provide a user analysis device. Refer to Figure 13 , Figure 13 which is the structural block diagram of a user analysis device 1300 provided by the embodiments of the present application. The user analysis device 1300 includes a first acquisition unit 1301, a second acquisition unit 1302, and a determination unit 1303:

[0109] The first acquisition unit 1301 is configured to acquire user behavior data corresponding to a target user, and the user behavior data is used to identify a target interaction object associated with the target user;

[0110] The second acquisition unit 1302 is configured to acquire an interaction object vector corresponding to the target interaction object. The interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users in multiple interaction objects, and the multiple interaction objects include the target interaction object;

[0111] The determination unit 1303 is configured to determine a security parameter corresponding to the target user according to the interaction object vector, and the security parameter is used to identify the probability of the target user having a risk interaction behavior.

[0112] In a possible implementation manner, the second acquisition unit 1302 is specifically configured to:

[0113] Generate an interaction object network based on the historical user behavior data. The interaction object network includes interaction object nodes corresponding to the multiple interaction objects respectively, and directed connections for connecting the interaction object nodes. The directed connections are generated based on the interaction order of the multiple users among the multiple interaction objects, and the directed connections are used to reflect the interaction weights between the connected interaction object nodes.

[0114] Determine interaction object vectors corresponding to the multiple interaction objects respectively according to the interaction object network. The multiple interaction objects include the target interaction object.

[0115] Obtain the interaction object vector corresponding to the target interaction object.

[0116] In a possible implementation manner, the second obtaining unit 1302 is specifically configured to:

[0117] Set a random walk parameter for executing a random walk strategy.

[0118] Generate a set of interaction object sequences according to the interaction object network through the random walk strategy of setting the random walk parameter. The set of interaction object sequences includes multiple interaction object sequences.

[0119] Based on the set of interaction object sequences, determine interaction object vectors corresponding to the multiple interaction objects respectively through a word vector model.

[0120] In a possible implementation manner, the user behavior data is the user behavior data corresponding to the target user in a target time interval, and the apparatus 1300 further includes a third obtaining unit:

[0121] The third obtaining unit is configured to obtain interaction data corresponding to the target user in the target time interval. The interaction data is used to identify any one or a combination of the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user.

[0122] The determining unit 1303 is specifically configured to:

[0123] Determine a security parameter corresponding to the target user according to the interaction object vector and the interaction data.

[0124] In a possible implementation manner, the determining unit 1303 is specifically configured to:

[0125] Determine the interaction time interval and the interaction amount corresponding to the target object according to the interaction data. The interaction time interval is used to identify the time interval for the target user to interact between adjacent interaction objects in time sequence, and the interaction amount is the interaction amount corresponding to the target interaction object.

[0126] Normalize the interaction time interval and the interaction amount;

[0127] Concatenate the normalized interaction time interval and the normalized interaction amount with the interaction object vector corresponding to the target interaction object to generate a fused interaction vector;

[0128] Determine the security parameter corresponding to the target user based on the fused interaction vector.

[0129] In a possible implementation manner, the determining unit 1303 is specifically configured to:

[0130] Determine the security parameter corresponding to the target user based on the fused interaction vector through a time series model.

[0131] The embodiments of the present application further provide a computer device, which will be introduced below with reference to the accompanying drawings. Please refer to Figure 14 As shown, the embodiments of the present application provide a device, which may also be a terminal device. The terminal device may be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA for short), a point of sales (POS for short), an in-vehicle computer, etc. Taking the terminal device as a mobile phone as an example:

[0132] Figure 14 The block diagram of a part of the structure of the mobile phone related to the terminal device provided by the embodiments of the present application is shown. Refer to Figure 14 , the mobile phone includes: a radio frequency (RF) circuit 1410, a memory 1420, an input unit 1430, a display unit 1440, a sensor 1450, an audio circuit 1460, a wireless fidelity (WiFi) module 1470, a processor 1480, and a power supply 1490, etc. Those skilled in the art can understand that Figure 8 the structure of the mobile phone shown in

[0133] does not limit the mobile phone, and may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements. Figure 14 The following will specifically introduce each component of the mobile phone:

[0134] The RF circuit 1410 can be used for receiving and transmitting information or signals during a call. Specifically, after receiving the downlink information from the base station, it is sent to the processor 1480 for processing. Additionally, the uplink data designed is transmitted to the base station. Generally, the RF circuit 1410 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF circuit 1410 can also communicate with the network and other devices through wireless communication. The above wireless communication can use any communication standard or protocol, including but not limited to the Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

[0135] The memory 1420 can be used to store software programs and modules. The processor 1480 executes various functional applications and data processing of the mobile phone by running the software programs and modules stored in the memory 1420. The memory 1420 mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory 1420 can include a high-speed random access memory and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0136] The input unit 1430 can be used to receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the mobile phone. Specifically, the input unit 1430 can include a touch panel 1431 and other input devices 1432. The touch panel 1431, also known as a touch screen, can collect touch operations of the user thereon or nearby (such as operations of the user using any suitable object or accessory such as a finger, a stylus, etc. on or near the touch panel 1431), and drive corresponding connection devices according to a preset program. Optionally, the touch panel 1431 can include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the touch orientation of the user, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 1480, and can receive and execute commands sent by the processor 1480. In addition, the touch panel 1431 can be implemented in multiple types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 1431, the input unit 1430 can also include other input devices 1432. Specifically, the other input devices 1432 can include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control keys, power on / off keys, etc.), a trackball, a mouse, a joystick, etc.

[0137] The display unit 1440 can be used to display information input by the user or information provided to the user and various menus of the mobile phone. The display unit 1440 can include a display panel 1441. Optionally, the display panel 1441 can be configured in forms such as a liquid crystal display (LCD) and an organic light-emitting diode (OLED). Further, the touch panel 1431 can cover the display panel 1441. When the touch panel 1431 detects a touch operation thereon or nearby, it transmits it to the processor 1480 to determine the type of touch event. Subsequently, the processor 1480 provides corresponding visual output on the display panel 1441 according to the type of touch event. Although in Figure 14 the touch panel 1431 and the display panel 1441 are implemented as two independent components to realize the input and input functions of the mobile phone, in some embodiments, the touch panel 1431 and the display panel 1441 can be integrated to realize the input and output functions of the mobile phone.

[0138] The mobile phone may further include at least one sensor 1450, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor. Among them, the ambient light sensor can adjust the brightness of the display panel 1441 according to the brightness of the ambient light, and the proximity sensor can turn off the display panel 1441 and / or the backlight when the mobile phone is moved to the ear. As a kind of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (generally three axes), and can detect the magnitude and direction of gravity when stationary, and can be used for applications that identify the posture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; as for other sensors that the mobile phone can also be configured with, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be elaborated here.

[0139] The audio circuit 1460, the speaker 1461, and the microphone 1462 can provide an audio interface between the user and the mobile phone. The audio circuit 1460 can transmit the electrical signal converted from the received audio data to the speaker 1461, and the speaker 1461 converts it into a sound signal for output; on the other hand, the microphone 1462 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1460 and then converted into audio data. After the audio data is output to the processor 1480 for processing, it is sent through the RF circuit 1410 to, for example, another mobile phone, or the audio data is output to the memory 1420 for further processing.

[0140] WiFi belongs to short - range wireless transmission technology. The mobile phone can help users send and receive emails, browse the web, and access streaming media through the WiFi module 1470, which provides users with wireless broadband Internet access. Although Figure 14 the WiFi module 1470 is shown, it can be understood that it does not belong to an essential component of the mobile phone and can be omitted completely within the scope of not changing the essence of the invention according to needs.

[0141] The processor 1480 is the control center of the mobile phone, connecting various parts of the entire mobile phone through various interfaces and circuits. By running or executing software programs and / or modules stored in the memory 1420, and by calling the data stored in the memory 1420, it executes various functions of the mobile phone and processes data. Optionally, the processor 1480 may include one or more processing units; preferably, the processor 1480 may integrate an application processor and a modem processor. Among them, the application processor mainly processes the operating system, user interface, and application programs, etc., and the modem processor mainly processes wireless communication. It can be understood that the above - mentioned modem processor may not be integrated into the processor 1480 either.

[0142] The mobile phone further includes a power source 1490 (such as a battery) for supplying power to each component. Preferably, the power source can be logically connected to the processor 1480 through a power management system, so as to implement functions such as management of charging, discharging, and power consumption management through the power management system.

[0143] Although not shown, the mobile phone may further include a camera, a Bluetooth module, etc., which will not be elaborated here.

[0144] In this embodiment, the processor 1480 included in the terminal device further has the following functions:

[0145] Obtain user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user;

[0146] Obtain an interaction object vector corresponding to the target interaction object, where the interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users among multiple interaction objects, and the multiple interaction objects include the target interaction object;

[0147] Determine a security parameter corresponding to the target user according to the interaction object vector, where the security parameter is used to identify the probability of the target user having a risk interaction behavior.

[0148] The embodiment of the present application further provides a server. Please refer to Figure 15 as shown Figure 15 is a structural diagram of the server 1500 provided by the embodiment of the present application. The server 1500 may vary greatly due to configuration or performance differences, and may include one or more central processing units (Central Processing Units, abbreviated as CPUs) 1522 (for example, one or more processors) and a memory 1532, and one or more storage media 1530 (for example, one or more mass storage devices) for storing application programs 1542 or data 1544. Among them, the memory 1532 and the storage media 1530 can be transient storage or persistent storage. The program stored in the storage media 1530 may include one or more modules (not marked in the figure), and each module may include a series of instruction operations on the server. Further, the central processor 1522 may be configured to communicate with the storage media 1530 and execute a series of instruction operations in the storage media 1530 on the server 1500.

[0149] The server 1500 may also include one or more power supplies 1526, one or more wired or wireless network interfaces 1550, one or more input / output interfaces 1558, and / or one or more operating systems 1541, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, and so on.

[0150] In the above embodiments, the steps performed by the server may be based on Figure 15 the server structure shown.

[0151] The embodiments of the present application also provide a computer-readable storage medium for storing a computer program, which is used to execute any one of the user analysis methods described in the foregoing various embodiments.

[0152] Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium can be at least one of the following media: read-only memory (abbreviation: ROM), RAM, magnetic disk, or optical disk, etc., which can store program codes.

[0153] It should be noted that the various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments. The device and system embodiments described above are only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0154] The above is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A user analysis method, characterized in that, The method includes: Obtaining user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user; Generating an interaction object network according to historical user behavior data. The interaction object network includes interaction object nodes respectively corresponding to multiple interaction objects and directed connections for connecting the interaction object nodes. The directed connections are generated based on the interaction order of multiple users among the multiple interaction objects, and the directed connections are used to reflect the interaction weights between the connected interaction object nodes; Determining interaction object vectors respectively corresponding to the multiple interaction objects according to the interaction object network, where the multiple interaction objects include the target interaction object; Obtaining the interaction object vector corresponding to the target interaction object, where the interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users among multiple interaction objects, and the multiple interaction objects include the target interaction object; Determining a security parameter corresponding to the target user according to the interaction object vector, where the security parameter is used to identify the probability of the target user having a risky interaction behavior.

2. The method according to claim 1, wherein The determining the interaction object vectors respectively corresponding to the multiple interaction objects according to the interaction object network includes: Setting a random walk parameter for executing a random walk strategy; Generating a set of interaction object sequences according to the interaction object network through the random walk strategy with the set random walk parameter, where the set of interaction object sequences includes multiple interaction object sequences; Determining the interaction object vectors respectively corresponding to the multiple interaction objects based on the set of interaction object sequences through a word vector model.

3. The method according to claim 1, characterized in that, The user behavior data is the user behavior data corresponding to the target user within a target time interval, and the method further includes: Obtaining interaction data corresponding to the target user within the target time interval, where the interaction data is used to identify any one or a combination of the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user; The determining the security parameter corresponding to the target user according to the interaction object vector includes: Determining the security parameter corresponding to the target user according to the interaction object vector and the interaction data.

4. The method according to claim 3, characterized in that If the interaction data is used to identify the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user, the determining the security parameter corresponding to the target user according to the interaction object vector and the interaction data includes: Determining the interaction time interval and the interaction amount corresponding to the target interaction object according to the interaction data, where the interaction time interval is used to identify the time interval for the target user to interact between adjacent interaction objects in time sequence, and the interaction amount is the interaction amount corresponding to the target interaction object; Performing normalization processing on the interaction time interval and the interaction amount; Concatenate the normalized interaction time interval and the normalized interaction amount with the interaction object vector corresponding to the target interaction object to generate a fused interaction vector; Determine the security parameter corresponding to the target user based on the fused interaction vector.

5. The method according to claim 4, characterized in that, The determining the security parameter corresponding to the target user based on the fused interaction vector includes: Determine the security parameter corresponding to the target user based on the fused interaction vector through a time series model.

6. A user analysis device, characterized in that, The apparatus includes a first acquisition unit, a second acquisition unit, and a determination unit: The first acquisition unit is configured to acquire user behavior data corresponding to a target user, where the user behavior data is used to identify a target interaction object associated with the target user; The second acquisition unit is configured to generate an interaction object network according to historical user behavior data. The interaction object network includes interaction object nodes corresponding to multiple interaction objects respectively, and directed connections for connecting the interaction object nodes. The directed connections are generated based on the interaction order of multiple users among the multiple interaction objects, and the directed connections are used to reflect the interaction weights between the connected interaction object nodes; According to the interaction object network, determine the interaction object vectors corresponding to the multiple interaction objects respectively, where the multiple interaction objects include the target interaction object; Obtain the interaction object vector corresponding to the target interaction object. The interaction object vector is generated based on historical user behavior data, and the historical user behavior data is used to reflect the interaction order of multiple users among multiple interaction objects, where the multiple interaction objects include the target interaction object; The determination unit is configured to determine the security parameter corresponding to the target user according to the interaction object vector, where the security parameter is used to identify the probability of the target user having a risky interaction behavior.

7. The device according to claim 6, characterized in that, The second acquisition unit is specifically configured to: Set a random walk parameter for executing a random walk strategy; Generate a set of interaction object sequences through the random walk strategy of setting the random walk parameter according to the interaction object network. The set of interaction object sequences includes multiple interaction object sequences; Based on the set of interaction object sequences, determine the interaction object vectors corresponding to the multiple interaction objects respectively through a word vector model.

8. The device according to claim 6, characterized in that, The user behavior data is the user behavior data corresponding to the target user within a target time interval, and the apparatus further includes a third acquisition unit; The third acquisition unit is configured to acquire interaction data corresponding to the target user within the target time interval, where the interaction data is used to identify any one or a combination of the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user; The determination unit is specifically configured to: Determine the security parameter corresponding to the target user according to the interaction object vector and the interaction data.

9. The device according to claim 8, wherein If the interaction data is used to identify the time intervals between multiple interaction behaviors corresponding to the target user and the interaction amounts of multiple interaction behaviors corresponding to the target user, the determination unit is specifically configured to: Determine the interaction time interval and interaction amount corresponding to the target interaction object according to the interaction data, where the interaction time interval is used to identify the time interval for the target user to interact between adjacent interaction objects in time sequence, and the interaction amount is the interaction amount corresponding to the target interaction object; Perform normalization processing on the interaction time interval and the interaction amount; Concatenate the normalized interaction time interval and the normalized interaction amount with the interaction object vector corresponding to the target interaction object to generate a fused interaction vector; Determine the security parameter corresponding to the target user based on the fused interaction vector.

10. The device according to claim 9, characterized in that, The determining unit is specifically configured to: Determine the security parameter corresponding to the target user based on the fused interaction vector through a time series model.

11. A computer device, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the user analysis method according to any one of claims 1-5 based on the instructions in the program code.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program, and the computer program is used to execute the user analysis method according to any one of claims 1-5.

Images