A Method and System for User Behavior Tendency Analysis Based on Multimodal Social Network Data

Abstract

This invention discloses a method and system for analyzing user behavior tendencies based on multimodal social network data, belonging to the field of user behavior analysis technology. Specifically, it discloses the classification of user social network data, determination of data types, and extraction of key factors through a key factor identification model to form key factor groups; based on the key factor groups, determination of corresponding user behavior tags to form user behavior tag groups; comparison of user behavior tag groups with preset user behavior comparison templates, and determination of initial predicted behavior tendencies based on the comparison results; and analysis of the repetition features of the initial predicted behavior tendencies of multiple sets of social network data to determine the user's actual behavior tendencies. This invention achieves accurate prediction of user behavior tendencies by systematically processing multimodal data, extracting key factors and constructing behavior tags, and combining template comparison and repetition feature analysis.

Description

Technical Field

[0001] This invention relates to the field of user behavior analysis technology, and in particular to a method and system for analyzing user behavior tendencies based on multimodal social network data. Background Technology

[0002] With the rapid development of digital technology, the data generated by users on various online platforms is becoming increasingly diverse and complex, encompassing multiple modalities such as text, images, audio and video, and interactive information. This multimodal data contains user behavioral characteristics and preference information, providing an important foundation for a deeper understanding of user intent. However, due to the heterogeneity and high dimensionality of multimodal data, traditional analysis methods are often limited to a single modality, making it difficult to comprehensively integrate cross-modal features, thus limiting the accuracy and breadth of behavioral preference analysis. Existing technologies typically face challenges in processing multimodal data, including unsystematic data classification, incomplete feature extraction, insufficient behavioral correlation modeling, and poor prediction robustness. In particular, how to extract key information from multi-source heterogeneous data, construct correlations between behavioral features, and achieve accurate behavioral preference prediction based on this remains a significant challenge in the technical field. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for analyzing user behavior tendencies.

[0004] This invention discloses a method for analyzing user behavior tendencies based on multimodal social network data, including:

[0005] Step S100: Classify the user's social network data, determine the network data type, and identify key factors of the social network data based on the key factor identification model corresponding to the network data type to obtain several key factors corresponding to a social network data, which are denoted as key factor groups.

[0006] Step S200: Based on the key factor group, determine several user behavior tags corresponding to the social network data to obtain the user behavior tag group;

[0007] Step S300: Compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, determine the user's initial predicted behavior tendency.

[0008] Step S400: Based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user, determine the user's actual behavioral tendencies.

[0009] In some embodiments disclosed in this invention, network data types include:

[0010] Text data, image data, video data, audio data, interactive data, metadata, and multimodal combined data.

[0011] In some embodiments disclosed in this invention, key factors include:

[0012] Key factors of text data include sentiment, keywords, semantic features, language style, text length, entities, and temporal dynamics.

[0013] Key factors of image data include objects, scenes, color features, facial features, image labels, image quality, and text overlay.

[0014] Key factors of video data include video content, temporal features, audio sentiment, subtitle text, video duration, interaction frequency, and shooting style;

[0015] Key factors in audio data include sentiment, tone, volume, speech rate, keywords, background noise, speech features, and duration.

[0016] Key factors of interaction data include interaction frequency, interaction objects, social network structure, interaction type, time pattern, and content preference;

[0017] Key factors of metadata include timestamp, geolocation, device information, network environment, platform type, and access frequency;

[0018] Key factors for multimodal composite data include cross-modal association features, sentiment association features, interaction effect features, intermodal weights, and contextual features.

[0019] In some embodiments disclosed in this invention, the method for determining several user behavior tags corresponding to social network data based on key factor sets includes:

[0020] Step S201: Construct a user behavior tag library, which includes several user behavior tags, and perform correlation analysis on the user behavior tags to construct several user behavior tag relationship expression models, wherein the tag relationship expression models are used to express the degree of correlation between user behavior tags;

[0021] Step S202: Compare the key factor group with the preset key factor-user behavior tag correspondence one by one. If the key factor group contains the key factor in the key factor-user behavior tag correspondence, extract the user behavior tag recorded in the key factor-user behavior tag correspondence and record it as the initial user behavior tag.

[0022] Step S203: Determine the user behavior label relationship expression model corresponding to the initial user behavior label, compare the determined user behavior label relationship expression models with each other, select a number of label relationship expression models with high degree of consistency, and identify the user behavior labels corresponding to the selected label relationship expression models as the user behavior labels corresponding to the social network data.

[0023] In some embodiments disclosed in this invention, the method for constructing several user behavior tag relationship expression models includes:

[0024] Step S2011: Each user behavior tag in the user behavior tag library is identified as a core user behavior tag. The data is then searched in historical social network data to identify other accompanying user behavior tags that appear together with the searched tags. The accompanying user behavior tags and the core user behavior tags are then combined to obtain a user behavior tag group.

[0025] Step S2012: Determine the frequency of occurrence of each accompanying user behavior tag in the user behavior tag group, and determine the degree of correlation between the accompanying user behavior tag and the core user behavior tag based on the frequency ratio of occurrence to the frequency of retrieved historical social network data.

[0026] Step S2013: Using the core user behavior tags as the model center point, establish a tag relevance expression line for the accompanying user behavior tags. A relevance mapping point is set on the tag relevance expression line. The distance between the relevance mapping point and the end of the relevance expression line is determined according to the relevance. The combination of several tag relevance expression lines and the model center point is identified as the user behavior tag relationship expression model.

[0027] In some embodiments disclosed in this invention, the method for comparing user behavior tag relationship expression models includes:

[0028] Step S2031: Randomly combine the user behavior label relationship expression models to obtain several user behavior label relationship model groups. Compare the user behavior label relationship models in the user behavior label relationship model groups to determine the degree of consistency between them, and calculate the average degree of consistency of all user behavior label relationship model groups.

[0029] Step S2032: Select user behavior tag relationship model groups with a matching degree greater than or equal to the average matching degree, and retain the core user behavior tags corresponding to the selected user behavior tag relationship model groups as several user behavior tags corresponding to social network data.

[0030] In some embodiments disclosed in this invention, the method for determining the degree of consistency between user behavior tag relationship models includes:

[0031] Step S20311: Compare the relative correlation lines between user behavior tag relationship models, determine the number of identical type pairs of the same type of correlation lines, and calculate the ratio of the number of identical type pairs to the total number of correlation lines.

[0032] Step S20312: Determine the relative distance between the points in the correlation mapping of each equivalent type relative performance line, and combine the equivalent type ratio to determine the degree of consistency between user behavior label relationship models.

[0033] In some embodiments disclosed in this invention, the method for comparing each user behavior tag group with a preset user behavior comparison template includes:

[0034] Step S301: Compare the user behavior tags in the user behavior tag group with the user behavior tags in the user behavior comparison template. If the equal ratio is greater than or equal to the preset value, the user behavior tendency corresponding to the user behavior comparison template is identified as the initial predicted behavior tendency.

[0035] In some embodiments disclosed in this invention, a method for determining a user's actual behavioral tendencies based on repetitive features of initial predicted behavioral tendencies corresponding to several sets of social network data includes:

[0036] Step S401: Determine whether the initial predicted behavioral tendencies belong to the same behavioral tendency category, retain the initial predicted behavioral tendencies that belong to the same behavioral tendency category, and calculate the repetition ratio of the retained initial predicted behavioral tendencies to all initial predicted behavioral tendencies. If the repetition ratio is greater than or equal to a preset value, then the retained initial predicted behavioral tendencies are identified as actual behavioral tendencies.

[0037] In some embodiments disclosed in this invention, a user behavior tendency analysis system based on multimodal social network data is also disclosed, including:

[0038] The first module is used to classify users' social network data, determine the network data type, and identify key factors of social network data based on the key factor identification model corresponding to the network data type, thereby obtaining several key factors corresponding to a social network data, which are denoted as key factor groups.

[0039] The second module is used to determine several user behavior tags corresponding to social network data based on key factor groups, and obtain user behavior tag groups.

[0040] The third module is used to compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, the user's initial predicted behavior tendency is determined.

[0041] The fourth module is used to determine the user's actual behavioral tendencies based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user.

[0042] This invention discloses a method and system for analyzing user behavior tendencies based on multimodal social network data, belonging to the field of user behavior analysis technology. Specifically, it discloses the classification of user social network data, the determination of data types (such as text, images, etc.), and the extraction of key factors through a key factor identification model to form key factor groups; based on the key factor groups, the determination of corresponding user behavior tags to form user behavior tag groups; the comparison of user behavior tag groups with preset user behavior comparison templates, and the determination of initial predicted behavior tendencies based on the comparison results; and the analysis of the repetition features of the initial predicted behavior tendencies of multiple sets of social network data to determine the user's actual behavior tendencies. This invention achieves accurate prediction of user behavior tendencies by systematically processing multimodal data, extracting key factors and constructing behavior tags, and combining template comparison and repetition feature analysis.

[0043] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the steps of the user behavior tendency analysis method based on multimodal social network data disclosed in this embodiment of the invention. Detailed Implementation

[0045] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0046] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. It should be understood that the preferred embodiments described herein are only for illustration and explanation of the present invention and should not be construed as limiting the scope of protection of the present invention. Those skilled in the art can make some non-essential improvements and adjustments based on the following content of the present invention. In the present invention, unless otherwise expressly specified and limited, the technical terms used in the present invention should have the ordinary meaning understood by those skilled in the art.

[0047] Example:

[0048] This invention discloses a method for analyzing user behavior tendencies based on multimodal social network data. (See attached document.) Figure 1 ,include:

[0049] Step S100: Classify the user's social network data, determine the network data type, and identify key factors of the social network data based on the key factor identification model corresponding to the network data type, thereby obtaining several key factors corresponding to a social network data, which are denoted as key factor groups.

[0050] Step S100 lays the foundation for subsequent behavioral tendency analysis by systematically classifying and extracting key factors from user social network data. First, social network data is categorized into various types, such as text, images, video, audio, interactive data, metadata, and multimodal combined data, each with unique characteristics and information expression methods. For example, text data may include user-posted posts, and image data may include user-uploaded photos. Based on the differences in data types, the system employs corresponding key factor identification models to selectively extract factors that reflect user behavioral characteristics. For example, for text data, key factors include sentiment, keywords, and semantic features; for image data, they include objects, scenes, and facial features. Through these models, the system extracts several key factors from complex multimodal data, forming key factor groups. For example, a user-posted image with text might have a key factor group including the keyword "travel" in the text and the "beach" scene in the image. These key factor groups provide a structured feature foundation for the subsequent generation of behavioral tags, ensuring that the analysis process covers the diversity of multimodal data.

[0051] Step S200: Based on the key factor group, determine several user behavior labels corresponding to the social network data to obtain the user behavior label group.

[0052] Step S200 works by mapping key factor groups to user behavior tags, transforming the features of multimodal data into quantifiable behavioral descriptions. In this step, the system first constructs a user behavior tag library containing various behavioral tags, such as "positive sharing" and "socially active," and analyzes the correlation between tags using historical data to build a tag relationship expression model. Next, the system compares the key factor groups with a pre-defined "key factor-user behavior tag" correspondence to extract matching initial behavioral tags. For example, if the key factor group contains "positive sentiment" and "high interaction frequency," it might map to the "positive social interaction" tag. Subsequently, the system uses the tag relationship expression model to filter out behavioral tags that are highly correlated with other tags, ensuring the accuracy and consistency of the tag groups. For example, if a user's key factor group shows that their video content frequently involves "fitness" and has a high interaction frequency, it might generate the tags "fitness enthusiast" and "socially active." This process, through a structured tag generation mechanism, transforms multimodal features into explicit descriptions of user behavior, providing a basis for subsequent tendency prediction.

[0053] Step S300: Compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, determine the user's initial predicted behavior tendency.

[0054] The principle of step S300 is to preliminarily infer the user's behavioral tendencies by comparing the user behavior tag group with the preset template. The system matches the tags in the user behavior tag group with the preset user behavior comparison template one by one. Each template corresponds to a behavioral tendency, such as "consumption tendency" or "information sharing tendency".

[0055] Step S400: Based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user, determine the user's actual behavioral tendencies.

[0056] The principle of step S400 is to optimize and determine the user's actual behavioral tendencies by analyzing the recurring characteristics of the initial predicted behavioral tendencies corresponding to multiple sets of social network data. The system collects multiple sets of social network data from users at different times or in different scenarios, generates initial predicted behavioral tendencies for each, and then determines whether these tendencies belong to the same behavioral tendency category. For example, if multiple predictions show "high consumption tendency," they are classified into the same category. The system calculates the percentage of these recurring tendencies; if the percentage exceeds a preset threshold (e.g., 70%), the tendency is confirmed as an actual behavioral tendency.

[0057] In some embodiments disclosed in this invention, network data types include:

[0058] Text data, image data, video data, audio data, interactive data, metadata, and multimodal combined data.

[0059] In some embodiments disclosed in this invention, key factors include:

[0060] Key factors of text data include sentiment, keywords, semantic features, language style, text length, entities, and temporal dynamics.

[0061] Key factors in image data include objects, scenes, color features, facial features, image labels, image quality, and text overlay.

[0062] Key factors of video data include video content, temporal features, audio sentiment, subtitle text, video duration, interaction frequency, and shooting style.

[0063] Key factors in audio data include emotional tone, tone, volume, speech rate, keywords, background noise, speech features, and duration.

[0064] Key factors of interaction data include interaction frequency, interaction objects, social network structure, interaction type, time pattern, and content preference.

[0065] Key factors of metadata include timestamps, geographic location, device information, network environment, platform type, and access frequency.

[0066] Key factors for multimodal composite data include cross-modal association features, sentiment association features, interaction effect features, intermodal weights, and contextual features.

[0067] In some embodiments disclosed in this invention, the method for determining several user behavior tags corresponding to social network data based on key factor sets includes:

[0068] Step S201: Construct a user behavior tag library, which includes several user behavior tags. Perform correlation analysis on the user behavior tags and construct several user behavior tag relationship expression models, wherein the tag relationship expression models are used to express the degree of correlation between user behavior tags.

[0069] The principle behind step S201 is to construct a user behavior tag library and establish a correlation model between tags through correlation analysis, providing a foundation for the accurate generation of subsequent behavior tags. First, the system creates a tag library containing various user behavior tags, such as "active sharing," "fitness enthusiast," and "high spending tendency," which reflect typical user behavior patterns on social networks. Next, the system performs correlation analysis on the tags, mining co-occurrence relationships between tags through historical social network data to construct a user behavior tag relationship expression model.

[0070] Step S202: Compare the key factor group with the preset key factor-user behavior tag correspondence one by one. If the key factor group contains the key factor in the key factor-user behavior tag correspondence, extract the user behavior tag recorded in the key factor-user behavior tag correspondence and record it as the initial user behavior tag.

[0071] Step S202 works by comparing key factor groups with preset mapping relationships to initially extract user behavior tags related to social network data. The system matches key factor groups (such as "positive sentiment" in text or "fitness scene" in images) one by one with the predefined "key factor-user behavior tag" correspondence. If a key factor group contains a key factor in the mapping relationship, the corresponding user behavior tag is extracted as the initial user behavior tag. For example, if a key factor group includes "fitness keywords" and "high interaction frequency," and the preset mapping relationship indicates that these factors correspond to the "fitness enthusiast" tag, then that tag is extracted as the initial tag. This process relies on the accuracy and comprehensiveness of the preset mapping relationship, which is usually constructed based on historical data or expert knowledge.

[0072] Step S203: Determine the user behavior label relationship expression model corresponding to the initial user behavior label, compare the determined user behavior label relationship expression models with each other, select a number of label relationship expression models with high degree of consistency, and identify the user behavior labels corresponding to the selected label relationship expression models as the user behavior labels corresponding to the social network data.

[0073] Step S203 works by comparing the tag relationship representation models corresponding to the initial user behavior tags to filter out highly consistent behavior tags, ensuring the accuracy and relevance of the final tags. The system first determines the tag relationship representation model corresponding to each initial user behavior tag (e.g., "fitness enthusiast"), which describes the degree of relevance of the tag to other tags. Next, the system compares these models and calculates the degree of consistency between them (e.g., through similarity or relevance scores of tag association lines). Models with high consistency are selected, and their corresponding user behavior tags are identified as the final user behavior tags for the social network data.

[0074] In some embodiments disclosed in this invention, the method for constructing several user behavior tag relationship expression models includes:

[0075] Step S2011: Each user behavior tag in the user behavior tag library is identified as a core user behavior tag. The data is then searched in historical social network data to identify other accompanying user behavior tags that appear together with the searched tags. The accompanying user behavior tags and the core user behavior tags are then combined to obtain a user behavior tag group.

[0076] Step S2011 works by analyzing historical social network data to identify accompanying tags that co-occur with core user behavior tags, constructing user behavior tag groups containing related tags to provide a foundation for subsequent correlation analysis. The system first identifies each tag in the user behavior tag library (e.g., "fitness enthusiast") as a core user behavior tag. Then, it retrieves other tags that co-occur with the core tag in historical social network data; these are called accompanying user behavior tags. For example, if "fitness enthusiast" frequently appears in user data alongside tags like "actively sharing" and "healthy lifestyle," these tags are identified as accompanying tags. Core tags and accompanying tags combine to form user behavior tag groups, such as {"fitness enthusiast," "actively sharing," "healthy lifestyle"}. This process relies on the richness of historical data and the statistical regularity of tag co-occurrence. For example, a user who frequently posts fitness-related posts and actively interacts may have data containing both the tags "fitness enthusiast" and "socially active." By constructing tag groups, the system captures the potential relationships between tags, providing a structured data foundation for subsequent quantitative correlation analysis.

[0077] Step S2012: Determine the frequency of occurrence of each accompanying user behavior tag in the user behavior tag group, and determine the correlation between the accompanying user behavior tag and the core user behavior tag based on the ratio of the occurrence frequency to the frequency of the retrieved historical social network data.

[0078] Step S2013: Using the core user behavior tags as the model center point, establish a tag relevance expression line for the accompanying user behavior tags. A relevance mapping point is set on the tag relevance expression line. The distance between the relevance mapping point and the end of the relevance expression line is determined according to the relevance. The combination of several tag relevance expression lines and the model center point is identified as the user behavior tag relationship expression model.

[0079] Step S2013 works by constructing a tag relevance expression line centered on the core user behavior tag, forming a user behavior tag relationship expression model that structurally expresses the associations between tags. The system uses the core tag (e.g., "fitness enthusiast") as the model's center point and establishes a tag relevance expression line for each accompanying tag (e.g., "active sharing" and "healthy living"). Relevance mapping points are set on the expression line, and their distance from the end of the expression line is determined by the relevance calculated in step S2012; the higher the relevance, the closer the mapping point is to the center point. For example, if the relevance between "active sharing" and "fitness enthusiast" is 0.75, their mapping point is closer, while the mapping point for "leisure and entertainment," with a relevance of 0.3, is farther away. Ultimately, the combination of the core tag and multiple expression lines constitutes a tag relationship expression model, resembling a radial diagram centered on the core tag.

[0080] In some embodiments disclosed in this invention, the method for comparing user behavior tag relationship expression models includes:

[0081] Step S2031: Randomly combine the user behavior label relationship expression models to obtain several user behavior label relationship model groups. Compare the user behavior label relationship models in the user behavior label relationship model groups to determine the degree of consistency between them, and calculate the average degree of consistency of all user behavior label relationship model groups.

[0082] Step S2032: Select user behavior tag relationship model groups with a matching degree greater than or equal to the average matching degree, and retain the core user behavior tags corresponding to the selected user behavior tag relationship model groups as several user behavior tags corresponding to social network data.

[0083] In some embodiments disclosed in this invention, the method for determining the degree of consistency between user behavior tag relationship models includes:

[0084] Step S20311: Compare the relative correlation lines between user behavior tag relationship models, determine the number of identical type pairs of the same type of correlation lines, and calculate the ratio of the number of identical type pairs to the total number of correlation lines.

[0085] Step S20312: Determine the relative distance between the points in the correlation mapping of each equivalent type relative performance line, and combine the equivalent type ratio to determine the degree of consistency between user behavior label relationship models.

[0086] The expression for calculating the degree of consistency is as follows:

[0087]

[0088] Where F represents the degree of similarity, H represents the ratio of identical types, and M represents the degree of similarity. max ΔM is the preset maximum relative distance between points. i Let be the relative distance between points of the i-th pair of related performance lines of the same type, and n be the total number of pairs of related performance lines of the same type.

[0089] In some embodiments disclosed in this invention, the method for comparing each user behavior tag group with a preset user behavior comparison template includes:

[0090] Step S301: Compare the user behavior tags in the user behavior tag group with the user behavior tags in the user behavior comparison template. If the equal ratio is greater than or equal to the preset value, the user behavior tendency corresponding to the user behavior comparison template is identified as the initial predicted behavior tendency.

[0091] In some embodiments disclosed in this invention, a method for determining a user's actual behavioral tendencies based on repetitive features of initial predicted behavioral tendencies corresponding to several sets of social network data includes:

[0092] Step S401: Determine whether the initial predicted behavioral tendencies belong to the same behavioral tendency category, retain the initial predicted behavioral tendencies that belong to the same behavioral tendency category, and calculate the repetition ratio of the retained initial predicted behavioral tendencies to all initial predicted behavioral tendencies. If the repetition ratio is greater than or equal to a preset value, then the retained initial predicted behavioral tendencies are identified as actual behavioral tendencies.

[0093] In some embodiments disclosed in this invention, a user behavior tendency analysis system based on multimodal social network data is also disclosed, including:

[0094] The first module is used to classify users' social network data, determine the network data type, and identify key factors of social network data based on the key factor identification model corresponding to the network data type, thereby obtaining several key factors corresponding to a social network data, which are denoted as key factor groups.

[0095] The second module is used to determine several user behavior tags corresponding to social network data based on key factor groups, and obtain user behavior tag groups.

[0096] The third module is used to compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, the user's initial predicted behavior tendency is determined.

[0097] The fourth module is used to determine the user's actual behavioral tendencies based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user.

[0098] This invention discloses a method and system for analyzing user behavior tendencies based on multimodal social network data, belonging to the field of user behavior analysis technology. Specifically, it discloses the classification of user social network data, determination of data types, and extraction of key factors through a key factor identification model to form key factor groups; based on the key factor groups, determination of corresponding user behavior tags to form user behavior tag groups; comparison of user behavior tag groups with preset user behavior comparison templates, and determination of initial predicted behavior tendencies based on the comparison results; and analysis of the repetition features of the initial predicted behavior tendencies of multiple sets of social network data to determine the user's actual behavior tendencies. This invention achieves accurate prediction of user behavior tendencies by systematically processing multimodal data, extracting key factors and constructing behavior tags, and combining template comparison and repetition feature analysis.

[0099] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for analyzing user behavior tendencies based on multimodal social network data, characterized in that, include: Step S100: Classify the user's social network data, determine the network data type, and identify key factors of the social network data based on the key factor identification model corresponding to the network data type to obtain several key factors corresponding to a social network data, which are denoted as key factor groups. Step S200: Based on the key factor group, determine several user behavior tags corresponding to the social network data to obtain the user behavior tag group; Step S300: Compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, determine the user's initial predicted behavior tendency. Step S400: Based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user, determine the user's actual behavioral tendencies. Methods for determining several user behavior tags corresponding to social network data based on key factor sets include: Step S201: Construct a user behavior tag library, which includes several user behavior tags, and perform correlation analysis on the user behavior tags to construct several user behavior tag relationship expression models, wherein the tag relationship expression models are used to express the degree of correlation between user behavior tags; Step S202: Compare the key factor group with the preset key factor ~ user behavior tag correspondence one by one. If the key factor group contains the key factor in the key factor ~ user behavior tag correspondence, extract the user behavior tag recorded in the key factor ~ user behavior tag correspondence and record it as the initial user behavior tag. Step S203: Determine the user behavior label relationship expression model corresponding to the initial user behavior label, compare the determined user behavior label relationship expression models with each other, select a number of label relationship expression models with high degree of consistency, and identify the user behavior labels corresponding to the selected label relationship expression models as the user behavior labels corresponding to the social network data. Methods for constructing several user behavior tag relationship representation models include: Step S2011: Each user behavior tag in the user behavior tag library is identified as a core user behavior tag. The data is then searched in historical social network data to identify other accompanying user behavior tags that appear together with the searched tags. The accompanying user behavior tags and the core user behavior tags are then combined to obtain a user behavior tag group. Step S2012: Determine the frequency of occurrence of each accompanying user behavior tag in the user behavior tag group, and determine the degree of correlation between the accompanying user behavior tag and the core user behavior tag based on the frequency ratio of occurrence to the frequency of retrieved historical social network data. Step S2013: Using the core user behavior tags as the model center point, establish a tag relevance expression line for the accompanying user behavior tags. A relevance mapping point is set on the tag relevance expression line. The distance between the relevance mapping point and the end of the relevance expression line is determined according to the relevance. The combination of several tag relevance expression lines and the model center point is identified as the user behavior tag relationship expression model.

2. The user behavior tendency analysis method based on multimodal social network data according to claim 1, characterized in that, Network data types include: Text data, image data, video data, audio data, interactive data, metadata, and multimodal combined data.

3. The user behavior tendency analysis method based on multimodal social network data according to claim 2, characterized in that, Key factors include: Key factors of text data include sentiment, keywords, semantic features, language style, text length, entities, and temporal dynamics. Key factors of image data include objects, scenes, color features, facial features, image labels, image quality, and text overlay. Key factors of video data include video content, temporal features, audio sentiment, subtitle text, video duration, interaction frequency, and shooting style; Key factors in audio data include sentiment, tone, volume, speech rate, keywords, background noise, speech features, and duration. Key factors of interaction data include interaction frequency, interaction objects, social network structure, interaction type, time pattern, and content preference; Key factors of metadata include timestamp, geolocation, device information, network environment, platform type, and access frequency; Key factors for multimodal composite data include cross-modal association features, sentiment association features, interaction effect features, intermodal weights, and contextual features.

4. The user behavior tendency analysis method based on multimodal social network data according to claim 1, characterized in that, Methods for comparing user behavior label relationship expression models include: Step S2031: Randomly combine the user behavior label relationship expression models to obtain several user behavior label relationship model groups. Compare the user behavior label relationship models in the user behavior label relationship model groups to determine the degree of consistency between them, and calculate the average degree of consistency of all user behavior label relationship model groups. Step S2032: Select user behavior tag relationship model groups with a matching degree greater than or equal to the average matching degree, and retain the core user behavior tags corresponding to the selected user behavior tag relationship model groups as several user behavior tags corresponding to social network data.

5. The user behavior tendency analysis method based on multimodal social network data according to claim 4, characterized in that, Methods for determining the degree of consistency between user behavior label relationship models include: Step S20311: Compare the relative correlation lines between user behavior tag relationship models, determine the number of identical type pairs of the same type of correlation lines, and calculate the ratio of the number of identical type pairs to the total number of correlation lines. Step S20312: Determine the relative distance between the points in the correlation mapping of each equivalent type relative performance line, and combine the equivalent type ratio to determine the degree of consistency between user behavior label relationship models.

6. The user behavior tendency analysis method based on multimodal social network data according to claim 1, characterized in that, The methods for comparing each user behavior tag group with the preset user behavior comparison template include: Step S301: Compare the user behavior tags in the user behavior tag group with the user behavior tags in the user behavior comparison template. If the equal ratio is greater than or equal to the preset value, the user behavior tendency corresponding to the user behavior comparison template is identified as the initial predicted behavior tendency.

7. The user behavior tendency analysis method based on multimodal social network data according to claim 1, characterized in that, Methods for determining a user's actual behavioral tendencies based on the repetitive features of initial predicted behavioral tendencies corresponding to several sets of social network data include: Step S401: Determine whether the initial predicted behavioral tendencies belong to the same behavioral tendency category, retain the initial predicted behavioral tendencies that belong to the same behavioral tendency category, and calculate the repetition ratio of the retained initial predicted behavioral tendencies to all initial predicted behavioral tendencies. If the repetition ratio is greater than or equal to a preset value, then the retained initial predicted behavioral tendencies are identified as actual behavioral tendencies.

8. A user behavior tendency analysis system based on multimodal social network data, characterized in that, The method for performing user behavior tendency analysis according to any one of claims 1-7 includes: The first module is used to classify users' social network data, determine the network data type, and identify key factors of social network data based on the key factor identification model corresponding to the network data type, thereby obtaining several key factors corresponding to a social network data, which are denoted as key factor groups. The second module is used to determine several user behavior tags corresponding to social network data based on key factor groups, and obtain user behavior tag groups. The third module is used to compare each user behavior tag group with a preset user behavior comparison template. Each user behavior comparison template corresponds to a user behavior tendency. Based on the template comparison, the user's initial predicted behavior tendency is determined. The fourth module is used to determine the user's actual behavioral tendencies based on the repetitive features of the initial predicted behavioral tendencies corresponding to several sets of social network data of the user.

Images