User interaction behavior prediction method and device and computer equipment

Abstract

The invention relates to a user interaction behavior prediction method and device, computer equipment, a storage medium and a computer program product. Relates to the technical field of big data. The method comprises the following steps: acquiring historical interaction behavior data of a user; performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; generating an interest file of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest file; and inputting the user portrait into the trained user behavior prediction model to obtain an interaction behavior prediction result of the user. By adopting the method, the data analysis efficiency of the social media user interaction behaviors can be improved.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a method, apparatus, computer equipment, storage medium and computer program product for predicting user interaction behavior. Background Art

[0002] With the advantages of content generation and information dissemination speed, social media has become an information sharing and online interaction tool used by more and more people every day. Users are the core of social networks, and user behavior is the starting point for understanding the operating mechanism of social networks. Therefore, the analysis of user data can reveal the user preferences and behavior patterns behind user online behavior and provide richer support for personalized recommendations on social media.

[0003] However, current user data analysis methods include page ranking, random walk, and iterative reduction decision tree algorithms, which have the problems of slow calculation speed, high resource consumption, and inability to effectively handle complex relationships between tags. Therefore, there is currently a problem of low efficiency in the process of data analysis of social media user interaction behavior. Summary of the invention

[0004] Based on this, it is necessary to provide a method, apparatus, computer device, computer-readable storage medium and computer program product for predicting user interaction behavior in order to address the technical problem of low efficiency in the process of performing data analysis on the above-mentioned social media user interaction behavior.

[0005] In a first aspect, the present application provides a method for predicting user interaction behavior, comprising:

[0006] Obtain the user's historical interaction behavior data;

[0007] Performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process;

[0008] generating an interest profile of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest profile;

[0009] The user portrait is input into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user.

[0010] In one of the embodiments, the historical interaction behavior data is subjected to fuzzy clustering processing to obtain category distribution information corresponding to each data point in the historical interaction behavior data, including: initializing a fuzzy membership matrix including each data point; each element of the fuzzy membership matrix represents the membership of the data point to each category; for each data point, obtaining the distance from the data point to a plurality of pre-set clustering centers, and updating the fuzzy membership matrix according to the distance; when the change of the membership matrix is ​​lower than a pre-set threshold or reaches a maximum number of iterations, obtaining the category distribution information corresponding to each data point.

[0011] In one of the embodiments, after the historical interaction behavior data is subjected to fuzzy clustering processing to obtain the category distribution information corresponding to each data point, the method further includes: for each data point, comparing the membership of the data point to each of the categories with a preset membership threshold of the target category; when the membership corresponding to any category is greater than the preset membership threshold, taking any category as the target category corresponding to the user's behavior; and generating the user's interest file according to the target category.

[0012] In one of the embodiments, the trained user behavior prediction model is trained in the following manner: obtaining a user portrait of a sample user; performing feature extraction on the user portrait to obtain interactive behavior features of the sample user; and training an initial user behavior prediction model using the interactive behavior features as input variables to obtain the trained user behavior prediction model.

[0013] In one of the embodiments, obtaining the user portrait of the user based on the interest file includes: obtaining the short-term interests and long-term interests in the interest file; representing the short-term interests and long-term interests as short-term behavior vectors and long-term behavior vectors respectively according to the behavior intensity of the user in different categories in different time intervals; and obtaining the user portrait of the user based on the short-term behavior vectors and the long-term behavior vectors.

[0014] In one of the embodiments, the fuzzy clustering processing of the historical interaction behavior data includes: deduplication, denoising and formatting the historical interaction behavior data to obtain cleaned data; converting the cleaned data into a uniform lowercase form, removing stop words and extracting keywords.

[0015] In one of the embodiments, after inputting the user portrait into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user, it includes: displaying the interactive behavior prediction result on a social media platform, and generating a recommendation list for the user based on the interactive behavior prediction result.

[0016] In a second aspect, the present application also provides a device for predicting user interaction behavior, comprising:

[0017] Data acquisition module, used to obtain historical interaction behavior data of users;

[0018] A data processing module, configured to perform fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the degree of membership of at least one category of each data point, and the categories represent different groups in the fuzzy clustering processing process;

[0019] A user portrait generation module, used to generate an interest file of the user according to the category distribution information, and obtain a user portrait of the user based on the interest file;

[0020] The prediction result acquisition module is used to input the user portrait into the trained user behavior prediction model to obtain the user's interactive behavior prediction result.

[0021] In a third aspect, the present application further provides a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the following steps when executing the computer program:

[0022] Obtain the user's historical interaction behavior data;

[0023] Performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process;

[0024] generating an interest profile of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest profile;

[0025] The user portrait is input into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user.

[0026] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0027] Obtain the user's historical interaction behavior data;

[0028] Performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process;

[0029] generating an interest profile of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest profile;

[0030] The user portrait is input into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user.

[0031] In a fifth aspect, the present application further provides a computer program product, the computer program product comprising a computer program, which implements the following steps when executed by a processor:

[0032] Obtain the user's historical interaction behavior data;

[0033] Performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process;

[0034] generating an interest profile of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest profile;

[0035] The user portrait is input into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user.

[0036] The above-mentioned prediction method, device, computer equipment, storage medium and computer program product for user interaction behavior have the following beneficial effects in the process of predicting user interaction behavior: first, the user's historical interaction behavior data is obtained, which can truly reflect the user's behavior and preference, and provide data support for the subsequent prediction process; then the historical interaction behavior data is fuzzy clustered to obtain the category distribution information corresponding to each data point in the historical interaction behavior data. The uncertainty and ambiguity in the historical interaction behavior data can be reduced by fuzzy clustering, so that the clustering results of the historical interaction behavior data are more reliable and stable; wherein the category distribution information includes the membership of each data point under at least one category, and the category represents different groups in the fuzzy clustering process; then the user's interest file is generated according to the category distribution information, and the user portrait of the user is obtained based on the interest file. The interest file can help build a personalized recommendation strategy for the user; then the user portrait is input into the trained user behavior prediction model to obtain the user's interaction behavior prediction result. By inputting rich and detailed user portraits, the accuracy of the behavior prediction model can be improved, thereby improving the credibility of the prediction result; therefore, through the combined effect of the above process, the efficiency of data analysis of social media user interaction behavior can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 is a flow chart of a method for predicting user interaction behavior in one embodiment;

[0039] Figure 2 It is a schematic diagram of a process of performing fuzzy clustering processing on historical interaction behavior data in one embodiment;

[0040] Figure 3 is a flowchart of a step of predicting a user interaction behavior in one embodiment;

[0041] Figure 4 A schematic diagram of a specific flow chart of a user interaction behavior prediction process in one embodiment;

[0042] Figure 5 is a structural block diagram of a device for predicting user interaction behavior in one embodiment;

[0043] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0045] The present application embodiment provides a method for predicting user interaction behavior, which can be applied to Figure 1 In the application environment shown.

[0046] In one embodiment, Figure 1 As shown, a method for predicting user interaction behavior is provided. This embodiment uses the method applied to a terminal as an example for illustration. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps S102 to S106. Among them:

[0047] Step S102, obtaining the user's historical interaction behavior data.

[0048] Among them, historical interaction behavior data is the activity record on the platform or application.

[0049] Optionally, historical interaction behavior data may include clicks, comments, likes, and shares, all of which are data reflecting user preferences and engagement.

[0050] Step S104, fuzzy clustering processing is performed on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering process.

[0051] Among them, fuzzy clustering is a clustering analysis technique that allows data points to belong to multiple categories, not just one specific category; category distribution information is the degree of membership of each data point between different categories in the result obtained by fuzzy clustering processing; membership is used to indicate the degree to which an element belongs to a certain set, that is, the degree to which a data point belongs to a certain category.

[0052] Specifically, fuzzy clustering can assign a membership degree to each data point, indicating the degree to which it belongs to each category, and has a better processing effect on data with low certainty; category distribution information can be used to reflect the diversity of data points; the value of the membership degree usually ranges from 0 to 1, and the larger the value, the higher the degree to which the element belongs to the category. Correspondingly, the larger the value, the higher the degree to which the data point belongs to a certain category.

[0053] Step S106, generating a user's interest profile according to the category distribution information, and obtaining a user portrait of the user based on the interest profile.

[0054] Among them, the interest file is a file or data structure generated based on the user's historical interaction behavior and fuzzy clustering results, which is used to record the user's main interests and preferences; the user portrait is a comprehensive description of the user's characteristics, usually including interest preferences and behavior patterns.

[0055] Specifically, the interest file may include data in a vector, a list or other forms, and may also include the behavior weights of the user in different categories, thereby reflecting the user's interest preferences.

[0056] Step S108, input the user portrait into the trained user behavior prediction model to obtain the user's interactive behavior prediction result.

[0057] Among them, the interactive behavior prediction result is the output of the user behavior prediction model, which describes the user's possible future interactive behavior, and may include predicted user behaviors such as clicks, comments, and purchases.

[0058] In the above-mentioned method for predicting user interaction behavior, the user's historical interaction behavior data is first obtained, which can truly reflect the user's behavior and preference and provide data support for the subsequent prediction process; then the historical interaction behavior data is fuzzy clustered to obtain the category distribution information corresponding to each data point in the historical interaction behavior data. Fuzzy clustering can reduce the uncertainty and ambiguity in the historical interaction behavior data, making the clustering results of the historical interaction behavior data more reliable and stable; wherein, the category distribution information includes the membership of at least one category of each data point, and the category represents different groups in the fuzzy clustering process; then, the user's interest file is generated according to the category distribution information, and the user's user portrait is obtained based on the interest file. The interest file can help build a personalized recommendation strategy for the user; then the user portrait is input into the trained user behavior prediction model to obtain the user's interaction behavior prediction result. By inputting rich and detailed user portraits, the accuracy of the behavior prediction model can be improved, thereby improving the credibility of the prediction result; therefore, through the combined effect of the above process, the efficiency of data analysis of social media user interaction behavior can be improved.

[0059] In an exemplary embodiment, Figure 3 As shown, step S104 performs fuzzy clustering processing on the historical interaction behavior data to obtain the category distribution information corresponding to each data point in the historical interaction behavior data, including steps S202 to S206, wherein:

[0060] Step S202, initialize the fuzzy membership matrix including each data point; each element of the fuzzy membership matrix represents the membership of the data point to each category; step S204, for each data point, obtain the distance from the data point to a plurality of pre-set clustering centers, and update the fuzzy membership matrix according to the distance; step S206, when the change of the membership matrix is ​​lower than a pre-set threshold or reaches the maximum number of iterations, obtain the category distribution information corresponding to each data point.

[0061] Among them, the data point can represent the interactive behavior data. The cluster center is the central point representing the category in the fuzzy clustering, which is usually the average point or centroid of each category. It can be used to measure the similarity between the data point and each category, and the membership of the data point is determined by calculating the distance. The category distribution information is the final membership result of each data point in the fuzzy clustering, which indicates the degree of belonging of each data point to different categories.

[0062] Specifically, in cluster analysis, distance calculation is a way to measure the similarity between data points and cluster centers. Methods such as Euclidean distance and Manhattan distance can be used. The smaller the distance, the closer the data point is to the cluster center, which can affect the update of membership. According to the distance calculation results between the data point and the cluster center, the values ​​in the fuzzy membership matrix can be dynamically adjusted.

[0063] In this embodiment, by initializing the fuzzy membership matrix, iteratively updating the membership and cluster centers until the stopping condition is met, the category distribution information of the data points is finally obtained, so that the model can flexibly respond to complex data, capture the diversity of data, and thus improve the accuracy of cluster analysis.

[0064] Furthermore, in one embodiment, after fuzzy clustering is performed on the historical interaction behavior data to obtain the category distribution information corresponding to each data point, it includes: for each data point, comparing the data point's membership to each category with a preset target category membership threshold; when the membership corresponding to any category is greater than the preset membership threshold, taking any category as the target category corresponding to the user's behavior; and generating the user's interest file according to the target category.

[0065] Among them, the target category is the category in a group of categories that is considered to best represent the characteristics or behaviors of a data point. It is usually based on the comparison of membership and can be used to reflect the most significant interest tendency of the data point. The interest file is a file generated based on the user's main category, recording the user's interests and preferences.

[0066] Specifically, the membership threshold is a preset critical value used to determine whether a category can be considered as the main category. If the membership of a data point to a category exceeds the threshold, the category is regarded as the target category of the data point.

[0067] In this embodiment, by analyzing the category distribution information and membership, the target category of each data point is decided. By comparing the membership with the preset threshold, the user's main interest areas can be accurately identified, and then the user's interest file can be generated, which improves the accuracy of data analysis and the effectiveness of user personalized services, and promotes a more accurate recommendation algorithm.

[0068] In one embodiment, the trained user behavior prediction model is trained in the following manner: obtaining a user portrait of a sample user; performing feature extraction on the user portrait to obtain interactive behavior features of the sample user; using the interactive behavior features as input variables to train the initial user behavior prediction model to obtain a trained user behavior prediction model.

[0069] Among them, user portrait is a comprehensive description of the user, usually including the user's basic information, such as age, gender, geographic location, interests, hobbies, historical behavior and preferences; feature extraction is the process of extracting useful information from raw data, which is used to reduce data dimensions and highlight important features; interactive behavior features refer to the user's behavior patterns during the interaction with the product, including click-through rate and browsing time.

[0070] Specifically, in user profiling, feature extraction may include the extraction of interaction frequency and preferred product categories; user behavior prediction models can predict users' future behaviors, such as purchase intentions and churn risks.

[0071] In this embodiment, more accurate personalized recommendations can be achieved by establishing detailed user portraits and extracting interactive behavior features; interactive behavior features are used as input variables to make the user behavior prediction model more accurate; the model obtained through training can more effectively predict the user's future behavior, and thus can formulate more effective marketing strategies.

[0072] In one embodiment, obtaining a user portrait of a user based on an interest file includes: obtaining short-term interests and long-term interests in the interest file; representing the short-term interests and long-term interests as short-term behavior vectors and long-term behavior vectors respectively according to the behavior intensity of the user in different categories in different time intervals; and obtaining a user portrait of the user based on the short-term behavior vectors and the long-term behavior vectors.

[0073] Among them, short-term interests refer to the interests and preferences shown by users in the recent period of time, which usually reflect the current needs and trends of users; long-term interests refer to the interests and preferences shown by users over a long period of time, reflecting the basic needs and living habits of users; behavior intensity refers to the frequency and degree of user interaction with a certain category in a specific time period; short-term behavior vector is a vector representation formed by quantifying the user's behavior intensity in the short term; correspondingly, long-term behavior vector is a vector representation formed by quantifying the user's behavior intensity in the long term; time interval refers to the time range selected when analyzing user behavior.

[0074] Specifically, short-term interests may change rapidly due to seasonal events, promotions, or hot topics; long-term interests are usually stable and not easily affected by short-term factors; high-intensity behavior usually indicates that the user has a greater interest in the category, while low-intensity behavior may indicate weakening or shifting interest; time intervals can be days, weeks, months, etc.

[0075] In this embodiment, by distinguishing between short-term and long-term interests, the enterprise can have a deeper understanding of the user's behavior patterns and psychological needs. The user portrait constructed using short-term and long-term behavior vectors can not only enable the enterprise to achieve more accurate personalized recommendations, thereby improving user retention and satisfaction, but also provide the enterprise with the ability to dynamically respond to market changes.

[0076] In an exemplary embodiment, fuzzy clustering processing is performed on the historical interaction behavior data, including: deduplication, denoising and formatting the historical interaction behavior data to obtain cleaned data; converting the cleaned data into a uniform lowercase form, removing stop words and extracting keywords.

[0077] Specifically, historical interactive behavior data may contain identical data due to multiple records or incorrect input, which will interfere with the accuracy of the analysis. The deduplication process usually includes comparing the key attributes of each record and retaining unique items to ensure the uniqueness of the data; abnormal behaviors in user behavior data, such as invalid clicks and malformed data, are regarded as noise, and denoising can improve the quality of the data; formatting processing includes converting data into a unified and standard format; converting text data into a unified lowercase form helps eliminate data differences caused by inconsistent uppercase and lowercase letters; removing stop words can make the keyword extraction process more efficient; keyword extraction is to identify words or phrases from text data that are more important for analysis.

[0078] In this embodiment, high-quality cleaned data is obtained by deduplicating, denoising, and formatting the historical interaction behavior data. These data are further unified into lowercase, stop words are removed, and keywords are extracted, which provides a solid foundation for subsequent analysis, recommendation system, and user portrait construction, and enhances the reliability and availability of the historical interaction behavior data.

[0079] Furthermore, in one embodiment, after inputting the user portrait into the trained user behavior prediction model to obtain the user's interactive behavior prediction result, it includes: displaying the interactive behavior prediction result on the social media platform, and generating a recommendation list for the user based on the interactive behavior prediction result.

[0080] Among them, the interactive behavior prediction results are obtained by analyzing the user's past interaction data, which are used to predict the user's possible future behavior and interests. They usually reflect the user's preferences and can be used to make personalized recommendations; the recommendation list is a list of personalized products, content or services generated based on user data and behavior predictions.

[0081] In this embodiment, by displaying the prediction results on the social media platform, user engagement can be effectively enhanced, providing a better user experience, helping companies to establish closer connections with users in a data-driven manner, optimize marketing strategies, improve conversion rates, and build an effective content supply mechanism based on user interests.

[0082] The present application provides a method for predicting user interaction behavior. In order to better understand the process of the above-mentioned method for predicting user interaction behavior, combined with Figure 3 As shown, the specific process of a method for predicting user interaction behavior of the present application is described in detail below, including the following steps:

[0083] Step S302: collect and process historical interaction behavior data.

[0084] Among them, social media platforms generate a large amount of user behavior data every day, including users' chat records, likes, comments, shares and other interactive information. The above data is collected from social media platforms to distributed storage systems through multiple interfaces. During the data collection process, it is necessary to ensure the integrity and accuracy of the data to avoid loss or duplication.

[0085] Specifically, the Hadoop distributed framework is used to process historical interactive behavior data. Through the parallel processing of historical interactive behavior data, the speed and efficiency of big data analysis are significantly improved. The distributed computing method can efficiently process massive data from multiple servers. The fuzzy clustering algorithm maps data points to multiple labels through fuzzy membership functions, thereby achieving more accurate user behavior classification. Fuzzy clustering allows a data point to belong to multiple categories at the same time, which is suitable for processing complex and dynamic user behavior data in social media. Through clustering analysis of user historical behavior data, a user preference prediction model is constructed. This model not only takes into account the user's short-term behavior, such as recent chat and interaction records, but also combines long-term behavior, such as continuous attention points in the past period of time, so as to more accurately predict the user's future preferences. In addition, a modular design is adopted to modularize functions such as data preprocessing, clustering analysis, and preference prediction. Each module can be updated and expanded independently, which is convenient for continuous optimization and upgrading of the system.

[0086] Step S304, constructing an interest file based on the processed historical interaction behavior data and obtaining a user portrait.

[0087] The collected data usually contains noise, such as duplicate records, invalid characters, and null values. The purpose of data cleaning is to remove these noises and improve data quality. Specific operations include: deduplication: removing duplicate chat records and interactive data; denoising: removing irrelevant characters and punctuation marks, as well as possible advertising information; formatting: converting data from different sources into a unified structured format; and in order to eliminate the differences between different data sources, data needs to be standardized, such as converting text data into a unified lowercase form, removing stop words, extracting keywords, etc. The standardized data is stored in the Hadoop distributed file system HDFS (Hierarchical Distributed File System) to ensure efficient access to data between different nodes; before clustering, text data needs to be converted into numerical form, which is usually achieved through methods such as BoW (Bag of Words), word embedding, or TF-IDF (Term Frequency-Inverse Document Frequency). For example, each chat record is represented as a high-dimensional vector, and each dimension of the vector represents the weight of a word in the record.

[0088] Specifically, the specific steps of the fuzzy clustering algorithm are as follows: Initialization: distribute the data in the Hadoop cluster and initialize the fuzzy membership matrix of each data point. Assume that we divide the data into CCC categories, and each element of the matrix represents the membership of the data point to a certain category; Iterative update: Update the membership matrix by repeatedly calculating the distance from each data point to the cluster center. The update of the cluster center is based on weighted average calculation, taking into account the membership of all data points; Termination condition: When the change of the membership matrix is ​​lower than the set threshold or reaches the maximum number of iterations, the algorithm stops iterating and outputs the final clustering result; The clustering result includes the fuzzy membership of each data point and the corresponding category distribution. In order to facilitate subsequent processing, these results will be further refined into the main categories of user behavior, such as "emotional expression", "interest topics", "social interaction", etc.

[0089] Step S306: input the user portrait into a pre-trained user behavior prediction model to predict the user's behavior.

[0090] Among them, after cluster analysis of each user's chat records and interactive behaviors, a user interest profile will be generated, which comprehensively considers the user's short-term interests and long-term interests, represented by short-term behavior vectors and long-term behavior vectors respectively. The formula is as follows:

[0091] Short-term interest vector: M=[S1,S2,…,Sn]M=[S_1,S_2,\dots,S_n]M=[S1,S2,…,Sn], each SiS_iSi represents the intensity of the user’s recent behavior in a certain category.

[0092] Long-term interest vector: N=[L1,L2,…,Lm]N=[L_1,L_2,\dots,L_m]N=[L1,L2,…,Lm], each LjL_jLj represents the user’s continued attention to a certain category over a long period of time.

[0093] By combining short-term interest and long-term interest vectors, a complete user portrait is generated. The user portrait not only contains interest preferences, but also combines behavioral patterns, such as interaction frequency, social relationship strength, etc. This information will be used as input data for subsequent model training and personalized recommendations. Based on the user portrait data, a prediction model is trained to predict the user's future behavior or interest changes. Commonly used models include decision tree-based models, random forests, support vector machines, or neural network models.

[0094] Specifically, data set partitioning is to divide user portrait data into training sets and test sets to ensure the effectiveness and generalization ability of model training; feature extraction is to extract key features from user portraits, such as short-term interests, long-term interests, social relationships, etc., as input to the model; model selection and training is to select the most suitable machine learning model based on the characteristics of the data, and train it through cross-validation, hyperparameter optimization and other methods to improve the model's predictive performance; after training is completed, the model will be used to predict the user's future behavior, such as content that the user may be interested in, social activities that the user may participate in, etc. The prediction results will serve as the basis of a personalized recommendation system to guide content recommendations and advertising delivery.

[0095] Step S308: output the predicted user behavior through the data visualization module, and generate recommended content based on the predicted user behavior.

[0096] Among them, in order to make the prediction results more intuitive, the present invention integrates a data visualization module to display user behavior patterns and prediction results in the form of charts, heat maps, user behavior path maps, etc. The visualization results can help social media platform operators better understand user needs and make corresponding strategy adjustments; and the prediction results will be directly applied to the recommendation system of the social media platform. According to the user's interests and behavior predictions, a recommendation list is automatically generated to push content or advertisements that the user may be interested in.

[0097] In the above embodiment, if Figure 4As shown in the figure, it includes the data flow, processing process, execution subject and time arrangement process, mainly through data collection and preprocessing, fuzzy clustering analysis, user interest preference modeling, model training and prediction, result output and visualization, each step is implemented in a distributed computing environment to ensure efficient processing of large-scale social media data; by introducing the Hadoop distributed computing framework and using multi-node parallel processing of large-scale data, the speed of data processing has been significantly improved, especially in the scenarios of massive data analysis and real-time processing, the performance is particularly outstanding; through the improved fuzzy clustering algorithm, user behavior patterns can be more accurately identified, and a more comprehensive user preference model is constructed by combining short-term and long-term user data. The model shows higher accuracy in predicting future user behavior, and the error is significantly reduced in multi-label classification problems; by adopting a modular system architecture design, each functional module can be independently developed, deployed and maintained, which not only improves the scalability of the system, but also enables the system to quickly adapt to new needs and changing environments. For example, when a new social media platform is added, only the data preprocessing module needs to be adjusted without the need to adjust the entire system. The system has undergone large-scale modifications, which greatly improved the flexibility and adaptability of the system. Through in-depth analysis and precise modeling of user behavior data, the system can provide more accurate content recommendations based on the user's historical behavior and preferences, thereby improving user participation and satisfaction. Especially in the field of advertising, by accurately analyzing the user's behavior patterns and interests, it can help advertisers better locate the target audience, optimize the advertising effect, and thus improve the conversion rate of advertising. In addition, in addition to social media platforms, the technical solution of the present application can also be applied to other fields that require large-scale data processing, such as e-commerce, financial services, and healthcare. In these fields, through the analysis of user behavior or other types of data, it can help enterprises or institutions make more accurate decisions and improve business efficiency and service quality. For example, from the perspective of management security and compliance, not only should chat software be used to solve the problem of work convenience, but also it is necessary to ensure that employees meet relevant safety requirements during work, strictly control the scope of communication, comply with relevant regulations, and provide support for the company's compliance and legal inspections to avoid risks such as sensitive data dissemination and leakage.

[0098] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0099] Based on the same inventive concept, the embodiment of the present application also provides a prediction device for user interaction behavior for implementing the above-mentioned prediction method for user interaction behavior. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the one or more user interaction behavior prediction device embodiments provided below can refer to the limitations of the user interaction behavior prediction method above, and will not be repeated here.

[0100] In an exemplary embodiment, Figure 5 As shown, a prediction device for user interaction behavior is provided, including: a data acquisition module 501, a data processing module 502, a user portrait generation module 503 and a prediction result acquisition module 504, wherein:

[0101] Data acquisition module 501, used to acquire historical interaction behavior data of users;

[0102] The data processing module 502 is used to perform fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process;

[0103] A user portrait generation module 503 is used to generate a user interest file according to the category distribution information, and obtain a user portrait of the user based on the interest file;

[0104] The prediction result acquisition module 504 is used to input the user portrait into the trained user behavior prediction model to obtain the user's interactive behavior prediction result.

[0105] Furthermore, in one embodiment, the data processing module 502 is also used to initialize a fuzzy membership matrix including each data point; each element of the fuzzy membership matrix represents the membership of the data point to each category; for each data point, the distance from the data point to a plurality of pre-set clustering centers is obtained, and the fuzzy membership matrix is ​​updated according to the distance; when the change of the membership matrix is ​​lower than a pre-set threshold or reaches the maximum number of iterations, the category distribution information corresponding to each data point is obtained.

[0106] Furthermore, in one embodiment, the data processing module 502 is also used to analyze category distribution information to obtain the category membership corresponding to each data point; compare the category membership with a preset main category membership threshold, and when the category membership is greater than the preset main category membership threshold, use the category as the main category corresponding to the user's behavior; and generate the user's interest file based on the main category.

[0107] Furthermore, in one embodiment, the prediction result acquisition module 504 is also used to obtain a user portrait of a sample user; perform feature extraction on the user portrait to obtain interactive behavior features of the sample user; and train an initial user behavior prediction model using the interactive behavior features as input variables to obtain a trained user behavior prediction model.

[0108] Furthermore, in one embodiment, the user portrait generation module 503 is also used to obtain short-term interests and long-term interests in the interest file; according to the behavior intensity of the user in different categories in different time intervals, the short-term interests and long-term interests are represented as short-term behavior vectors and long-term behavior vectors respectively; based on the short-term behavior vectors and the long-term behavior vectors, the user portrait of the user is obtained.

[0109] Furthermore, in one embodiment, the data processing module 502 is also used to deduplicate, denoise and format the historical interaction behavior data to obtain cleaned data; convert the cleaned data into a uniform lowercase form, remove stop words and extract keywords.

[0110] Furthermore, in one embodiment, the prediction result acquisition module 504 is also used to display the interaction behavior prediction result on the social media platform and generate a user recommendation list based on the interaction behavior prediction result.

[0111] Each module in the above-mentioned user interaction behavior prediction device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0112] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 6 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store prediction data of user interaction behavior. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting user interaction behavior is implemented.

[0113] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0114] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0115] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0116] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0117] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0118] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0119] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for predicting user interaction behavior, characterized in that: The method comprises: Obtain the user's historical interaction behavior data; Performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the membership degree of each data point under at least one category, and the category represents different groups in the fuzzy clustering processing process; generating an interest profile of the user according to the category distribution information, and obtaining a user portrait of the user based on the interest profile; The user portrait is input into a trained user behavior prediction model to obtain the interactive behavior prediction result of the user.

2. The method according to claim 1, characterized in that The performing fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data includes: Initializing a fuzzy membership matrix including each data point; each element of the fuzzy membership matrix represents the membership of the data point to each category; For each data point, obtaining the distance from the data point to a plurality of pre-set cluster centers, and updating the fuzzy membership matrix according to the distance; When the change of the membership matrix is ​​lower than a preset threshold or reaches a maximum number of iterations, the category distribution information corresponding to each data point is obtained.

3. The method according to claim 2, characterized in that After the fuzzy clustering process is performed on the historical interaction behavior data to obtain the category distribution information corresponding to each data point, the method further includes: For each data point, compare the degree of membership of the data point to each of the categories with a preset threshold of membership of the target category, and when the degree of membership corresponding to any category is greater than the preset threshold of membership, use the any category as the target category corresponding to the user's behavior; Generate the user's interest file according to the target category.

4. The method according to claim 1, characterized in that: The trained user behavior prediction model is trained in the following way: Get user profiles of sample users; Extracting features from the user portrait to obtain interactive behavior features of the sample user; The initial user behavior prediction model is trained with the interactive behavior features as input variables to obtain the trained user behavior prediction model.

5. The method according to claim 1, characterized in that The obtaining a user portrait of the user based on the interest file includes: Obtaining short-term interests and long-term interests in the interest file; According to the behavior intensity of the user in different categories in different time intervals, the short-term interest and the long-term interest are represented as a short-term behavior vector and a long-term behavior vector respectively; A user portrait of the user is obtained based on the short-term behavior vector and the long-term behavior vector.

6. The method according to claim 1, characterized in that The performing fuzzy clustering processing on the historical interaction behavior data includes: De-duplication, de-noising and formatting of the historical interaction behavior data to obtain cleaned data; The cleaned data is converted into a uniform lowercase form, stop words are removed, and keywords are extracted.

7. The method according to claim 1, characterized in that After inputting the user portrait into the trained user behavior prediction model to obtain the interactive behavior prediction result of the user, the method includes: The interactive behavior prediction result is displayed on a social media platform, and a recommendation list for the user is generated based on the interactive behavior prediction result.

8. A device for predicting user interaction behavior, characterized in that: The device comprises: Data acquisition module, used to obtain historical interaction behavior data of users; A data processing module, configured to perform fuzzy clustering processing on the historical interaction behavior data to obtain category distribution information corresponding to each data point in the historical interaction behavior data; the category distribution information includes the degree of membership of each data point under at least one category, and the categories represent different groups in the fuzzy clustering processing process; A user portrait generation module, used to generate an interest file of the user according to the category distribution information, and obtain a user portrait of the user based on the interest file; The prediction result acquisition module is used to input the user portrait into the trained user behavior prediction model to obtain the user's interactive behavior prediction result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images