User portrait generation method and device, computer device, and storage medium

Abstract

The present application relates to the field of group portrait, disclose a kind of user portrait generation method, device, computer equipment and storage medium, its method includes: obtaining user interaction history record;According to user interaction history record generates user intent data, user intent data includes each user's flow node and node intent, a flow node corresponds to a node intent;According to user intent data constructs g×c two-dimensional matrix, wherein, g is total number of users, c is total number of node intent;Two-dimensional matrix is decomposed into first matrix and second matrix by non-negative matrix decomposition algorithm, first matrix is g×p matrix, second matrix is p×c matrix, and p is self-defined dimension;According to first matrix and second matrix, construct a number of group user portraits.The user portrait generated by the present application is highly matched with the intelligent customer service system, and the reply satisfaction of the intelligent customer service system can be improved.

Description

Technical Field

[0001] The present invention relates to the field of group portraits, and in particular to a method, apparatus, computer equipment and storage medium for generating user portraits. Background Art

[0002] Existing intelligent customer service systems generally generate responses based on guided conversational language. To simplify the design, responses to identical customer inquiries are typically identical. In some cases, user profiles can be introduced to meet personalized user needs and adjust responses based on these profiles to improve satisfaction with intelligent responses. However, existing user profiles are mostly generated based on third-party data collection and are poorly aligned with intelligent customer service systems. This results in a significant deviation between adjusted responses and actual user needs, leading to low satisfaction with intelligent responses. Summary of the Invention

[0003] Based on this, it is necessary to provide a user portrait generation method, device, computer equipment and storage medium to address the above technical problems, so as to generate a user portrait that is highly matched with the intelligent customer service system and improve the satisfaction with intelligent replies.

[0004] A method for generating a user portrait, comprising:

[0005] Get user interaction history;

[0006] Generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, where one process node corresponds to one node intention;

[0007] Constructing a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions;

[0008] Decomposing the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix, and the second matrix is ​​a p×c matrix, where p is a custom dimension;

[0009] Construct several group user portraits based on the first matrix and the second matrix.

[0010] A user portrait generating device, comprising:

[0011] Acquisition record module, used to obtain user interaction history;

[0012] An intention data generation module is used to generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, and each process node corresponds to a node intention;

[0013] A two-dimensional matrix construction module is used to construct a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions;

[0014] A matrix decomposition module, configured to decompose the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix and the second matrix is ​​a p×c matrix, where p is a custom dimension;

[0015] A group portrait construction module is used to construct several group user portraits based on the first matrix and the second matrix.

[0016] A computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the user portrait generation method described above is implemented.

[0017] One or more readable storage media storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to execute the user portrait generation method as described above.

[0018] The user profile generation method, apparatus, computer device, and storage medium described above acquire user interaction history records to identify user intent based on the user interaction history records. User intent data is generated based on the user interaction history records. The user intent data includes each user's process node and node intent, with each process node corresponding to a node intent. This generates user intent data related to the process node, and the generated user intent data is highly compatible with the intelligent customer service system. A g×c two-dimensional matrix is ​​constructed based on the user intent data, where g is the total number of users and c is the total number of node intents. This converts the user intent data into mathematical form, improving the processing capability of the user intent data. The two-dimensional matrix is ​​decomposed into a first matrix and a second matrix using a non-negative matrix factorization algorithm. The first matrix is ​​a g×p matrix, and the second matrix is ​​a p×c matrix, where p is a custom dimension. This reduces the dimensionality of the two-dimensional matrix, simultaneously decomposing the first matrix containing user statistical information and the second matrix containing user intent. Based on the first and second matrices, multiple group user profiles are constructed to characterize user profiles by group. The generated group user profiles include user intent data. The user profiles generated by the present invention are highly compatible with the intelligent customer service system, which can improve the response satisfaction of the intelligent customer service system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 This is a schematic diagram of an application environment of a method for generating a user portrait according to an embodiment of the present invention;

[0021] Figure 2 This is a flow chart of a method for generating a user portrait according to an embodiment of the present invention;

[0022] Figure 3 is a structural diagram of a user portrait generating device in one embodiment of the present invention;

[0023] Figure 4 FIG. 1 is a schematic diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0025] The user portrait generation method provided in this embodiment can be applied to Figure 1 In an application environment, a client communicates with a server. Clients include, but are not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0026] In one embodiment, if Figure 2 As shown, a user portrait generation method is provided, which is applied in Figure 1 The server in the example is used for explanation, and the steps include the following steps S10-S50.

[0027] S10. Obtain user interaction history records.

[0028] It is understood that the user interaction history can be a history of interactions between customer service personnel and users in the customer service system. The user interaction history can be text data or voice data. In some examples, the user interaction history relates to pre-sales consultation for a product. In other examples, the user interaction history relates to post-sales communication for a product.

[0029] S20. Generate user intention data based on the user interaction history records, where the user intention data includes a process node and node intention for each user, where one process node corresponds to one node intention.

[0030] Understandably, the user interaction history includes the conversation content of several process nodes. Process nodes can be set based on actual needs. In some examples, process nodes include but are not limited to self-introduction (customer service staff), product introduction, and abnormal order processing. Each process node can be pre-set with several optional intents. The conversation content of each process node can be converted into the node intent of the process node manually or machine-based. In one example, the optional intents of the product introduction node include interest and disinterest. If the user's reply is "This product is really good, I like it very much", the node intent of the node can be judged as "interested"; if the user's reply is "This product is average, I will take a look later", the node intent of the node can be judged as "disinterested". By processing the user interaction history, user intent data containing the user's process nodes and node intents can be generated.

[0031] S30. Construct a two-dimensional matrix of g×c according to the user intention data, where g is the total number of users and c is the total number of node intentions.

[0032] Understandably, when the total number of users is g and the total number of node intentions is c, the user intention data can be expressed as:

[0033] User 1: Node 11, intention 11; Node 12, intention 12; ...; Node 1c, intention Figure 1 c;

[0034] User 2: Node 21, intention 21; Node 22, intention 22; ...; Node 2c, intention Figure 2 c;

[0035] User 3: Node 31, intention 31; Node 32, intention 32; ...; Node 3c, intention Figure 3 c;

[0036] ...;

[0037] User g: node g1, intention g1; node g2, intention g2; ...; node gc, intention gc.

[0038] A two-dimensional matrix of g×c can be constructed based on the above user intention data.

[0039] S40. Decompose the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, where the first matrix is ​​a g×p matrix, the second matrix is ​​a p×c matrix, and p is a custom dimension.

[0040] It can be understood that the Nonnegative Matrix Factorization (NMF) algorithm is a multidimensional data processing method that can make all decomposed components non-negative and simultaneously achieve nonlinear dimensionality reduction.

[0041] The two-dimensional matrix is ​​represented by V, the first matrix decomposed is represented by W, and the second matrix is ​​represented by H. The non-negative matrix factorization algorithm can be expressed as: V = W·H + ε, where ε is the error term. The first matrix is ​​a g × p matrix, and the second matrix is ​​a p × c matrix, where p is a custom dimension. The value of p is determined by the actual matrix size. If the values ​​of g and c are relatively large, p can be smaller to reduce the amount of data calculations.

[0042] Under the constraint of non-negativity, the non-negative matrix factorization algorithm simplifies high-dimensional random patterns into low-dimensional random patterns H while keeping the information unchanged as much as possible. The basis of this simplification is to estimate the essential structure W in the data.

[0043] In the first matrix W, each row forms a vector representing a user's demographic information, such as age, height, and occupation. In the second matrix H, each row forms a vector representing an intent. The two-dimensional matrix V can be viewed as a matrix that combines user intent (H) and demographic information (W).

[0044] S50: Construct several group user portraits based on the first matrix and the second matrix.

[0045] Understandably, since the first matrix contains user statistics and the second matrix contains user intent, we can divide users into groups based on the first matrix, and then organize user intent by user group to build group user profiles.

[0046] In steps S10-S50, user interaction history records are obtained to identify user intent based on the user interaction history records. User intent data is generated based on the user interaction history records. The user intent data includes each user's process node and node intent, with each process node corresponding to a node intent. This generates user intent data related to the process node, and the generated user intent data is highly compatible with the intelligent customer service system. A g×c two-dimensional matrix is ​​constructed based on the user intent data, where g is the total number of users and c is the total number of node intents. This converts the user intent data into a mathematical form and improves the processing capabilities of the user intent data. The two-dimensional matrix is ​​decomposed into a first matrix and a second matrix using a non-negative matrix factorization algorithm. The first matrix is ​​a g×p matrix, and the second matrix is ​​a p×c matrix, where p is a custom dimension. This reduces the dimensionality of the two-dimensional matrix and simultaneously decomposes the matrix into a first matrix containing user statistical information and a second matrix containing user intent. Based on the first and second matrices, several group user profiles are constructed to characterize user profiles by group. The generated group user profiles include user intent data.

[0047] Optionally, step S20, i.e., generating user intention data based on the user interaction history records, includes:

[0048] S201, dividing the user interaction history record into a plurality of node data based on process nodes;

[0049] S202, obtaining a recognition model corresponding to the node data;

[0050] S203: Identify the node data according to the recognition model to obtain the node intent of the node data.

[0051] Understandably, the user interaction history records include a large number of conversations (one question and one answer). The user interaction history records can be divided based on the correlation between the questions in the conversation and the process nodes to form a number of node data, and each node data includes a number of conversations. A simple recognition model can be set for each process node, and the recognition model can be constructed by a semantic recognition algorithm. The recognition model corresponding to the node data refers to the recognition model corresponding to the process node where the node data is located. For example, the process node where node data 1 is located is node 1, and the recognition model 1 of node 1 can be obtained. Recognition model 1 is the recognition model corresponding to node data 1.

[0052] The recognition model then identifies the node data and obtains the node intent. Typically, a process node contains relatively little information, so the recognition model can quickly identify the node intent. User intent data contains the user's node intent at the process node.

[0053] In steps S201-S203, the user interaction history is segmented into multiple node data based on process nodes to facilitate determining the intent of each node data. A recognition model corresponding to the node data is obtained to identify the intent of the node data using the recognition model, thereby improving recognition efficiency. The node data is identified based on the recognition model, and the node intent of the node data is obtained to obtain user intent data.

[0054] Optionally, step S40, i.e., decomposing the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, includes:

[0055] S401. Process the two-dimensional matrix using a non-negative matrix decomposition equation to generate the first matrix and the second matrix. The non-negative matrix decomposition equation includes:

[0056]

[0057] Among them, V ij is the value of the i-th row and j-th column of the two-dimensional matrix V;

[0058] W ik is the value of the i-th row and k-th column of the first matrix;

[0059] H kj is the value of the kth row and jth column of the second matrix;

[0060] ε ij V ij The error term.

[0061] In this embodiment, the first matrix and the second matrix can be solved by the above equations. p is a custom dimension and can be based on the size of the two-dimensional matrix. When the two-dimensional matrix is ​​large (i.e., the values ​​of g and c are large), p can take a smaller value, such as 5-10. When the two-dimensional matrix is ​​small (i.e., the values ​​of g and c are small), p can take a larger value, such as 10-30. ε ij V ij The error term, when solving, ε ij The smaller the value, the better.

[0062] Optionally, step S50, i.e., constructing a plurality of group user portraits according to the first matrix and the second matrix, includes:

[0063] S501: Determine a valid user group k based on the maximum value of the kth column of the first matrix;

[0064] S502: Generate a group user portrait of the valid user group k according to the values ​​of the k-th row of the second matrix.

[0065] It can be understood that the first matrix is ​​a g×p matrix, which contains the statistical data of the users, and the second matrix is ​​a p×c matrix, which contains the user intentions. The first matrix is ​​processed column by column. In a column of data, the user corresponding to the maximum value can be regarded as the most representative user in the column. Valid user groups can be divided accordingly. In one example, users (including users corresponding to the maximum value) who are close to the maximum value (a numerical range can be set, and those within the numerical range are regarded as close to the maximum value) can be included in the valid user group. In another example, users corresponding to values ​​that are not close to the maximum value can also be divided into one or more sets according to the numerical distribution to form new user groups. Each user group can be associated with specific labels, and these labels belong to different personal information attributes. For example, labels include children, youth, tall people, etc., among which the labels children and youth belong to age attributes, and tall people belong to height attributes.

[0066] Since the first matrix has p columns, p valid user groups can be generated. The second matrix has p rows, each of which can be considered the user intent of a valid user group. That is, valid user group k, determined based on the maximum value in column k of the first matrix, has a group user profile that includes the value in row k of the second matrix. Therefore, the group user profile can be represented by a vector of length c. In some cases, the user intent in the second matrix can be adaptively modified by combining the intent data of users within the valid user groups in the two-dimensional matrix V.

[0067] In steps S501-S502, valid user group k is determined based on the maximum value in the kth column of the first matrix to distinguish a highly significant group. A group user profile for valid user group k is generated based on the value in the kth row of the second matrix, generating a group user profile that has a high degree of compatibility with the intelligent customer service system.

[0068] Optionally, after step S50, that is, after constructing a plurality of group user portraits according to the first matrix and the second matrix, the method further includes:

[0069] S51. Obtaining user's personal information;

[0070] S52: Obtain at least one group user portrait that matches the personal information;

[0071] S53: Generate a personal user portrait of the user based on the at least one group user portrait.

[0072] Understandably, a user's personal information includes, but is not limited to, age, gender, height, and occupation. Different features in personal information can be matched to different group user profiles. For example, if user A is young and tall, they can be matched to both the youth profile and the tall profile. A user's personal profile can be the weighted sum of all their group user profiles.

[0073] Optionally, after step S53, that is, after generating the individual user portrait of the user based on the at least one group user portrait, the method further includes:

[0074] S54, receiving user feedback information at a specified process node;

[0075] S55: Determine reply information for the user feedback information based on the personal user portrait.

[0076] It is understandable that the generated personal user portrait can be applied to the intelligent customer service system. The designated process node can be a pre-set node. For example, a designated process node is a product introduction node, which is used to provide product introductions to users. User feedback information refers to the feedback information provided by the user after receiving the product introduction. The above-mentioned user feedback information can be analyzed based on the personal user portrait of the user to determine the user's intention, and the adapted reply information can be selected or synthesized based on the user's intention. In some examples, several pieces of information can be pre-set, and different user intentions match different information. Therefore, after determining the user's intention, the information matching the user's intention can be determined as the reply information. In another example, a language (sound) synthesis model can be constructed. After determining the user's intention, a set of language (sound) parameters matching the user's intention can be obtained, and then the corresponding language (sound) reply information can be synthesized through the language (sound) synthesis model.

[0077] According to measured data, after using the user portrait generation method provided in this embodiment, the response satisfaction of the intelligent customer service system is increased by more than 10% compared with the original response satisfaction.

[0078] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0079] In one embodiment, a user portrait generation device is provided, which corresponds to the user portrait generation method in the above embodiment. Figure 3 As shown, the user portrait generation device includes a record acquisition module 10, an intention data generation module 20, a two-dimensional matrix construction module 30, a matrix decomposition module 40 and a group portrait construction module 50. The functional modules are described in detail as follows:

[0080] The record acquisition module 10 is used to acquire the user interaction history records;

[0081] An intention data generating module 20 is configured to generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, where one process node corresponds to one node intention;

[0082] A two-dimensional matrix construction module 30 is used to construct a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions;

[0083] A matrix decomposition module 40 is configured to decompose the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix and the second matrix is ​​a p×c matrix, where p is a custom dimension;

[0084] The group portrait building module 50 is used to build a plurality of group user portraits according to the first matrix and the second matrix.

[0085] Optionally, the intention data generating module 20 includes:

[0086] A data segmentation unit, configured to segment the user interaction history records into a plurality of node data based on process nodes;

[0087] An identification model acquisition unit is used to acquire an identification model corresponding to the node data;

[0088] The intention recognition unit is used to recognize the node data according to the recognition model and obtain the node intention of the node data.

[0089] Optionally, the matrix decomposition module 40 includes:

[0090] A matrix decomposition unit is used to process the two-dimensional matrix through a non-negative matrix decomposition equation to generate the first matrix and the second matrix, wherein the non-negative matrix decomposition equation includes:

[0091]

[0092] Among them, V ij is the value of the i-th row and j-th column of the two-dimensional matrix V;

[0093] W ik is the value of the i-th row and k-th column of the first matrix;

[0094] H kj is the value of the kth row and jth column of the second matrix;

[0095] ε ij V ij The error term.

[0096] Optionally, the group portrait building module 50 includes:

[0097] Determine a valid user group k based on the maximum value of the kth column of the first matrix;

[0098] A group user portrait of the effective user group k is generated according to the values ​​of the k-th row of the second matrix.

[0099] Optionally, the user portrait generating device further includes:

[0100] The module for obtaining personal information is used to obtain the user's personal information;

[0101] A portrait matching module, configured to obtain at least one group user portrait that matches the personal information;

[0102] The personal user portrait generation module is used to generate a personal user portrait of the user based on the at least one group user portrait.

[0103] Optionally, the user portrait generating device further includes:

[0104] A feedback information receiving module is used to receive user feedback information at a specified process node;

[0105] The reply information determination module is used to determine reply information for the user feedback information based on the personal user portrait.

[0106] For the specific definition of the user portrait generation device, please refer to the definition of the user portrait generation method above, which will not be repeated here. The various modules in the above-mentioned user portrait generation device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0107] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium and an internal memory. The readable storage medium stores an operating system, computer-readable instructions and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The database of the computer device is used to store data involved in the user portrait generation method. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer-readable instructions are executed by the processor, a user portrait generation method is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0108] In one embodiment, a computer device is provided, comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the following steps are implemented:

[0109] Get user interaction history;

[0110] Generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, where one process node corresponds to one node intention;

[0111] Constructing a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions;

[0112] Decomposing the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix, and the second matrix is ​​a p×c matrix, where p is a custom dimension;

[0113] Construct several group user portraits based on the first matrix and the second matrix.

[0114] In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided. The computer-readable storage media provided in this embodiment include non-volatile computer-readable storage media and volatile computer-readable storage media. The computer-readable storage media store computer-readable instructions that, when executed by one or more processors, implement the following steps:

[0115] Get user interaction history;

[0116] Generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, where one process node corresponds to one node intention;

[0117] Constructing a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions;

[0118] Decomposing the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix, and the second matrix is ​​a p×c matrix, where p is a custom dimension;

[0119] Construct several group user portraits based on the first matrix and the second matrix.

[0120] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through computer-readable instructions. The computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0121] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0122] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for generating a user portrait, characterized in that: include: Get user interaction history; Generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, where one process node corresponds to one node intention; Constructing a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions; Decomposing the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix, the second matrix is ​​a p×c matrix, and p is a custom dimension; Constructing a plurality of group user portraits according to the first matrix and the second matrix; The constructing a plurality of group user portraits according to the first matrix and the second matrix includes: Determine a valid user group k based on the maximum value of the kth column of the first matrix; Generate a group user portrait of valid user group k according to the value of the kth row of the second matrix; The users corresponding to values ​​that are not close to the maximum value are divided into one or more sets according to the numerical distribution to form a new user group, and the new user group is associated with a specific tag.

2. The user portrait generation method according to claim 1, wherein: Generating user intention data according to the user interaction history records includes: Segmenting the user interaction history records into a plurality of node data based on process nodes; Obtaining a recognition model corresponding to the node data; The node data is identified according to the identification model to obtain the node intent of the node data.

3. The user portrait generation method according to claim 1, wherein: Decomposing the two-dimensional matrix into a first matrix and a second matrix by a non-negative matrix decomposition algorithm includes: The two-dimensional matrix is ​​processed by a non-negative matrix decomposition equation to generate the first matrix and the second matrix, wherein the non-negative matrix decomposition equation includes: in, is the value of the i-th row and j-th column of the two-dimensional matrix V; is the value of the i-th row and k-th column of the first matrix; is the value of the kth row and jth column of the second matrix; for The error term.

4. The user portrait generation method according to claim 1, wherein: After constructing a plurality of group user portraits according to the first matrix and the second matrix, the method further includes: Obtaining users' personal information; Obtaining at least one group user portrait that matches the personal information; Generate a personal user portrait of the user based on the at least one group user portrait.

5. The user portrait generation method according to claim 4, characterized in that: After generating the individual user portrait of the user according to the at least one group user portrait, the method further includes: Receive user feedback information at specified process nodes; Determine reply information for the user feedback information based on the personal user portrait.

6. A user portrait generation device, characterized in that: include: Acquisition record module, used to obtain user interaction history; An intention data generation module is used to generate user intention data based on the user interaction history records, wherein the user intention data includes a process node and a node intention of each user, and each process node corresponds to a node intention; A two-dimensional matrix construction module is used to construct a g×c two-dimensional matrix based on the user intention data, where g is the total number of users and c is the total number of node intentions; A matrix decomposition module, configured to decompose the two-dimensional matrix into a first matrix and a second matrix using a non-negative matrix decomposition algorithm, wherein the first matrix is ​​a g×p matrix and the second matrix is ​​a p×c matrix, where p is a custom dimension; A group portrait construction module, configured to construct a plurality of group user portraits based on the first matrix and the second matrix; The constructing a plurality of group user portraits according to the first matrix and the second matrix includes: Determine a valid user group k based on the maximum value of the kth column of the first matrix; Generate a group user portrait of valid user group k according to the value of the k-th row of the second matrix; The users corresponding to values ​​that are not close to the maximum value are divided into one or more sets according to the numerical distribution to form a new user group, and the new user group is associated with a specific tag.

7. The user portrait generation device according to claim 6, wherein: The generating intention data module includes: A data segmentation unit, configured to segment the user interaction history records into a plurality of node data based on process nodes; An identification model acquisition unit is used to acquire an identification model corresponding to the node data; The intention recognition unit is used to recognize the node data according to the recognition model and obtain the node intention of the node data.

8. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein: When the processor executes the computer-readable instructions, the user portrait generation method according to any one of claims 1 to 5 is implemented.

9. One or more readable storage media storing computer-readable instructions, wherein when the computer-readable instructions are executed by one or more processors, the one or more processors execute the user portrait generation method according to any one of claims 1 to 5.

Images