Personalized Visual Communication Element Recommendation Method and System Based on Internet Big Data

Abstract

This invention relates to the field of visual element recommendation, and more particularly to a method and system for personalized visual communication element recommendation based on internet big data. The method includes the following steps: acquiring multi-dimensional internet behavior data of users and mining user interaction behavior on each platform to obtain a multi-platform user interaction content data pool; inferring user interaction intent from the multi-platform user interaction content data pool and mining personalized interaction interest preferences to obtain user interaction interest preferences; mining deep interaction needs based on user interaction interest preferences and performing adaptive profile evolution to construct a personalized user demand profile; and parsing visual elements and performing multimodal visual mining based on the multi-platform user interaction content data pool to obtain multimodal visual element information for each homepage. This invention improves the accuracy and quality of visual element recommendation through precise analysis of user aesthetic needs.

Description

Technical Field

[0001] This invention relates to the field of visual element recommendation, and more particularly to a personalized visual communication element recommendation method and system based on Internet big data. Background Technology

[0002] With the continuous development of information technology and digital media, visual communication, as the most direct and impactful form of human information interaction, plays an increasingly important role in various fields such as advertising and marketing, user interface design, and digital content production. Along with the continuous expansion of the internet user base and the rapid development of big data technology, user behavior data on internet platforms is becoming increasingly abundant, providing a solid data foundation for achieving more accurate and efficient personalized visual communication.

[0003] In today's internet environment, users engage in numerous interactions daily through various channels such as social media, e-commerce platforms, and content platforms. These interactions not only reflect users' content preferences and interests but also reveal diverse individual characteristics in areas such as visual aesthetics, cultural preferences, and interaction methods. Traditional visual communication methods often rely on human experience or generic templates for content design and presentation, lacking in-depth exploration and precise adaptation to individual user differences, thus failing to meet the growing demands of modern users for personalized expression and visual experiences.

[0004] Existing personalized recommendation methods primarily focus on recommending structured data such as text, products, or videos. However, they still face numerous challenges in recommending visual communication elements. On one hand, user visual preferences are highly subjective and dynamically changing, making accurate modeling difficult using single-dimensional data. On the other hand, existing systems lack the ability to integrate and process multimodal factors such as user aesthetic characteristics, cultural background, and interaction habits, resulting in recommendations that lack visual relevance and appeal. Furthermore, traditional methods mostly employ static rules or simple tag matching mechanisms, lacking a deep understanding of user behavior and the ability to intelligently generate visual elements, thus failing to achieve truly personalized visual communication. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a personalized visual communication element recommendation method and system based on Internet big data, thereby resolving at least one of the aforementioned technical problems.

[0006] To achieve the above objectives, this invention provides a personalized visual communication element recommendation method based on internet big data, comprising the following steps:

[0007] Step S1: Obtain multi-dimensional internet behavior data of users and mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool;

[0008] Step S2: Infer user interaction intent from the user's multi-platform interaction content data pool and mine personalized interaction interest preferences to obtain user interaction interest preferences;

[0009] Step S3: Based on user interaction interests and preferences, conduct in-depth interaction needs mining and adaptive profile evolution to build personalized user needs profiles;

[0010] Step S4: Perform visual element analysis and multimodal visual mining based on the user's multi-platform interaction content data pool to obtain multimodal visual element information for each homepage;

[0011] Step S5: Perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend;

[0012] Step S6: Based on the user's dynamic aesthetic trends, perform multi-visual element-driven learning on the personalized user demand profile, and perform feedback compensation optimization to build an intelligent visual element generation model.

[0013] This invention integrates user behavior data from multiple platforms (such as social media, e-commerce, content platforms, and search engines) to comprehensively reconstruct user behavior characteristics in different environments. Mining the interactive behavior characteristics of each platform helps form a well-structured and data-rich interactive content data pool, providing a solid foundation for subsequent preference inference and recommendation. Through inter-platform comparison and cross-validation, it identifies differentiated user behaviors across different platforms, forming a multi-dimensional behavior matrix, which is beneficial for accurately modeling users' true interests. Using intent recognition technologies (such as deep semantic analysis and contextual modeling), it infers users' true interaction motivations and goals from surface behavior. It identifies changes in user preferences at different times and in different contexts, dynamically depicting user interest maps. By integrating both intent and behavioral features, it avoids preference assumptions based solely on frequency, achieving a deeper understanding of user preferences and laying a precise interest-based foundation for subsequent recommendations. It extends from explicit interests to implicit needs (such as aesthetic trends, potential purchasing tendencies, and cognitive styles), uncovering users' latent needs. User profiles are not static; they evolve continuously in real time based on new user behaviors, ensuring that the profile always reflects the user's current state and adapts to fluctuations in their interests. This system constructs a demand profile across three layers: semantic, aesthetic, and behavioral, evolving from "user tags" to a "user cognitive model." It integrates visual elements such as images, text, typography, and animation to achieve a more realistic and comprehensive understanding of webpage / platform content. By combining visual elements with user semantic interests, it analyzes the potential correlations between patterns, colors, fonts, and user semantic preferences. Breaking through the limitations of traditional text recommendations, it incorporates visual communication into the system's analytical dimensions, making recommendations more intuitive and aesthetically pleasing. Based on users' cultural background, aesthetic preferences, and regional styles, it uncovers their deep-seated aesthetic and cultural foundations. By fitting user aesthetic trends, it achieves efficient matching between visual content and individual aesthetic sensibilities, making recommendation results more attractive. It tracks the trajectory of user aesthetic changes, achieving "aesthetic trend perception" in visual recommendations to meet users' expectations for visual updates and novelty. Based on models such as GANs (Generative Adversarial Networks), it generates visual communication elements that conform to user aesthetics and needs, breaking the limitations of fixed material libraries. Through user interaction feedback, it continuously optimizes recommended content, achieving visual co-creation between users and the system, enabling the system to continuously "learn" and better understand users. By using feedback mechanisms such as click-through rate, user rating, and dwell time to learn in reverse, the quality of visual element recommendations is continuously optimized, forming a self-closing loop.

[0014] This specification provides a personalized visual communication element recommendation system based on internet big data, used to execute the personalized visual communication element recommendation method based on internet big data as described above, including:

[0015] The behavior mining module is used to acquire multi-dimensional internet behavior data of users and to mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool.

[0016] The interest and preference module is used to infer user interaction intent from the user's multi-platform interaction content data pool and to mine personalized interaction interest preferences to obtain user interaction interest preferences.

[0017] The user profile module is used to deeply mine user interaction needs based on user interaction interests and preferences, and to adaptively evolve user profiles to build personalized user needs profiles.

[0018] The visual element parsing module is used to parse visual elements and mine multimodal visuals based on the user's multi-platform interaction content data pool to obtain multimodal visual element information for each homepage.

[0019] The aesthetic trend analysis module is used to perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend.

[0020] The feedback compensation and optimization module is used to learn multi-visual element-driven models of personalized user needs profiles based on the user's dynamic aesthetic trends, and to perform feedback compensation and optimization to build an intelligent visual element generation model.

[0021] This invention monitors user behavior across multiple internet platforms (such as social media, e-commerce websites, and news platforms) to obtain rich user interaction data, including clicks, browsing, comments, and sharing, laying a solid foundation for subsequent analysis. Mining user interactions across different platforms helps identify behavioral patterns and habits, understand their reactions in different contexts, and thus provide a basis for personalized recommendations. By analyzing the user interaction content data pool, it's possible to infer user interests and intentions, providing foundational data for personalized recommendation systems and helping the system more accurately understand user needs. Deeply mining user interaction interests and preferences helps platforms recommend content that better matches user interests, thereby improving user satisfaction and engagement. Through in-depth needs mining, dynamic personalized user profiles can be generated, reflecting users' latest needs and preferences. These profiles evolve with changes in user behavior, maintaining their timeliness and accuracy. The construction of personalized user needs profiles helps businesses develop more targeted strategies for product promotion and marketing, thereby improving conversion rates. Analyzing the visual elements of user interaction content yields rich multimodal visual information, including images, videos, and graphics, providing multi-dimensional data support for understanding user aesthetics and preferences. Multimodal visual mining enables recommendation systems to offer more diverse and richer visual content, enhancing the user experience. By analyzing users' aesthetic and cultural backgrounds, the platform can understand the aesthetic preferences of users from different cultural backgrounds, thus making recommendations more aligned with users' cultural identity. Constructing a dynamic aesthetic profile of users can reflect changes in their aesthetic needs in real time, helping the platform adjust recommendation strategies promptly and enhancing user experience. Based on the dynamic aesthetic profile of users, multi-visual element-driven learning can establish an intelligent visual element generation model. This model can automatically generate visual content that matches user preferences, improving the intelligence level of recommendations. A feedback compensation optimization mechanism can continuously improve recommendation performance based on actual user feedback, ensuring that the system's adaptability remains consistent with changes in user needs. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the steps of a personalized visual communication element recommendation method based on Internet big data according to the present invention.

[0023] Figure 2 This is a detailed flowchart illustrating the implementation steps of step S1.

[0024] Figure 3 This is a detailed flowchart illustrating the implementation steps of step S2;

[0025] Figure 4 This is a flowchart illustrating the detailed implementation steps of step S3. Detailed Implementation

[0026] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0027] This application provides a method and system for recommending personalized visual communication elements based on Internet big data. The executing entities of the method and system include, but are not limited to, mechanical equipment, data processing platforms, cloud server nodes, and network upload devices that can be considered general computing nodes in this application. The data processing platform includes, but is not limited to, at least one of an audio-visual management system, an information management system, and a cloud data management system.

[0028] Please see Figures 1 to 4 This invention provides a personalized visual communication element recommendation method based on Internet big data, which includes the following steps:

[0029] Step S1: Obtain multi-dimensional internet behavior data of users and mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool;

[0030] Step S2: Infer user interaction intent from the user's multi-platform interaction content data pool and mine personalized interaction interest preferences to obtain user interaction interest preferences;

[0031] Step S3: Based on user interaction interests and preferences, conduct in-depth interaction needs mining and adaptive profile evolution to build personalized user needs profiles;

[0032] Step S4: Perform visual element analysis and multimodal visual mining based on the user's multi-platform interaction content data pool to obtain multimodal visual element information for each homepage;

[0033] Step S5: Perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend;

[0034] Step S6: Based on the user's dynamic aesthetic trends, perform multi-visual element-driven learning on the personalized user demand profile, and perform feedback compensation optimization to build an intelligent visual element generation model.

[0035] This invention integrates user behavior data from multiple platforms (such as social media, e-commerce, content platforms, and search engines) to comprehensively reconstruct user behavior characteristics in different environments. Mining the interactive behavior characteristics of each platform helps form a well-structured and data-rich interactive content data pool, providing a solid foundation for subsequent preference inference and recommendation. Through inter-platform comparison and cross-validation, it identifies differentiated user behaviors across different platforms, forming a multi-dimensional behavior matrix, which is beneficial for accurately modeling users' true interests. Using intent recognition technologies (such as deep semantic analysis and contextual modeling), it infers users' true interaction motivations and goals from surface behavior. It identifies changes in user preferences at different times and in different contexts, dynamically depicting user interest maps. By integrating both intent and behavioral features, it avoids preference assumptions based solely on frequency, achieving a deeper understanding of user preferences and laying a precise interest-based foundation for subsequent recommendations. It extends from explicit interests to implicit needs (such as aesthetic trends, potential purchasing tendencies, and cognitive styles), uncovering users' latent needs. User profiles are not static; they evolve continuously in real time based on new user behaviors, ensuring that the profile always reflects the user's current state and adapts to fluctuations in their interests. This system constructs a demand profile across three layers: semantic, aesthetic, and behavioral, evolving from "user tags" to a "user cognitive model." It integrates visual elements such as images, text, typography, and animation to achieve a more realistic and comprehensive understanding of webpage / platform content. By combining visual elements with user semantic interests, it analyzes the potential correlations between patterns, colors, fonts, and user semantic preferences. Breaking through the limitations of traditional text recommendations, it incorporates visual communication into the system's analytical dimensions, making recommendations more intuitive and aesthetically pleasing. Based on users' cultural background, aesthetic preferences, and regional styles, it uncovers their deep-seated aesthetic and cultural foundations. By fitting user aesthetic trends, it achieves efficient matching between visual content and individual aesthetic sensibilities, making recommendation results more attractive. It tracks the trajectory of user aesthetic changes, achieving "aesthetic trend perception" in visual recommendations to meet users' expectations for visual updates and novelty. Based on models such as GANs (Generative Adversarial Networks), it generates visual communication elements that conform to user aesthetics and needs, breaking the limitations of fixed material libraries. Through user interaction feedback, it continuously optimizes recommended content, achieving visual co-creation between users and the system, enabling the system to continuously "learn" and better understand users. By using feedback mechanisms such as click-through rate, user rating, and dwell time to learn in reverse, the quality of visual element recommendations is continuously optimized, forming a self-closing loop.

[0036] In the embodiments of the present invention, see Figure 1 This is a flowchart illustrating the steps of a personalized visual communication element recommendation method based on internet big data according to the present invention. In this example, the steps of the personalized visual communication element recommendation method based on internet big data include:

[0037] Step S1: Obtain multi-dimensional internet behavior data of users and mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool;

[0038] This embodiment identifies multiple internet platforms to be analyzed, including social media (such as Facebook and Instagram), e-commerce (such as Amazon and eBay), and content sharing websites (such as YouTube and Pinterest). Data from these platforms will provide a comprehensive perspective for user behavior analysis. The characteristics of each platform are analyzed to determine the main types of user interaction behaviors (such as likes, comments, shares, and browsing), and the API interface documentation or data scraping strategies for each platform are recorded. The types of user behavior data to be collected are defined, such as user ID, behavior type, timestamp, and content type (images, videos, text). The goal is to collect comprehensive user behavior records on each platform for subsequent analysis. The APIs provided by each platform are used to access user behavior data. If a platform does not provide an API, web crawling techniques are used to scrape publicly available data. The terms of use for each platform must be followed to avoid data misuse violations. The database structure is designed to effectively store and manage the collected data. The database should contain all relevant information about user behavior records, including user ID, platform name, interaction time, and behavior type. The frequency of data scraping is set, such as daily scheduled scraping of user behavior data to ensure data timeliness. Plan the data scraping volume for each platform, for example, scraping 1000 of the latest user behavior records each time. Clean and preprocess the collected user behavior data to remove duplicate records and outliers. Ensure all data is consistent and accurate for subsequent analysis. Categorize user behavior data, identifying different user interaction types (such as likes, comments, shares, etc.), and assign weights to each behavior. This step helps in understanding user preferences. Set classification criteria, for example, assigning weight 1 to likes, weight 2 to comments, and weight 3 to shares, to better assess user engagement and preferences in subsequent analysis. Integrate the user behavior data collected from all platforms into a unified data pool, ensuring consistent data format for easy analysis; this step is the foundation for building multi-dimensional user data analysis. Index the integrated data pool to improve data query efficiency. Ensure that subsequent data analysis can quickly access data for specific users or specific behavior types.

[0039] Step S2: Infer user interaction intent from the user's multi-platform interaction content data pool and mine personalized interaction interest preferences to obtain user interaction interest preferences;

[0040] In this embodiment, a user interaction intent inference model is designed. This model should be able to analyze user interaction behavior across different platforms to identify their potential intent. Common methods include Natural Language Processing (NLP) techniques and machine learning algorithms (such as decision trees and random forests). User interaction behavior data, including comment content, number of likes, and sharing behavior, is extracted from a multi-platform user interaction content data pool. This data will serve as the model's input features. The diversity and representativeness of the feature data are ensured so that the model can accurately infer user intent. The input feature dimensions of the model are defined, such as: sentiment score of comment text, number of likes, number of shares, etc. An appropriate training set to test set ratio (e.g., 70% training set, 30% test set) is selected to ensure the model's effectiveness. Feature extraction is performed on user interaction behavior. Sentiment analysis tools are used to process comments, extracting sentiment scores and keywords. Simultaneously, user intent (e.g., information acquisition, product purchase, social interaction, etc.) is labeled according to the type of user behavior (e.g., comment, like). The labeled dataset is used to train the intent inference model. The model's performance is evaluated using cross-validation, and model parameters are optimized to improve the accuracy of inference. Multiple experimental parameters, such as learning rate and tree depth, can be set for parameter tuning. The inferred user intents are recorded in a database to ensure that each user's interaction behavior has a corresponding intent tag. The model's prediction results are analyzed to observe the distribution of different intent types in order to identify the user's primary intent. Based on the inferred user interaction intents, a personalized interaction interest preference analysis framework is constructed. This framework should be able to integrate the user's historical interaction behavior and inferred intents to extract the user's interest preferences. The user's intent inference results are integrated with interaction behavior data to analyze the user's preference for specific content or visual elements under different intents. For example, under the intent of "information acquisition," users may prefer articles and infographics, while under the intent of "social interaction," they may prefer images and videos. A threshold for preference extraction is set; if a user's interaction frequency with a certain type of content is more than 1.5 times the average, it is considered an interest preference for that content type. This method identifies the user's personalized interest domains.

[0041] Step S3: Based on user interaction interests and preferences, conduct in-depth interaction needs mining and adaptive profile evolution to build personalized user needs profiles;

[0042] This embodiment clarifies the goal of deep interaction need mining: identifying users' potential needs based on their interaction interests and preferences. These needs may include desires for specific products, services, content types, or experiences. Data related to user interaction interests and preferences is collected, including historical user behavior data, intent inference results, and sentiment analysis results. Data comprehensiveness is ensured so that the model can accurately identify user needs. A dataset partitioning ratio is set (e.g., 70% training set, 30% test set), and appropriate features are selected, including user interaction frequency, behavior type, and sentiment score, for subsequent model training. Feature extraction is performed on the collected data, and methods such as cluster analysis and association rule mining are used to identify users' deep interaction needs. By analyzing users' high-frequency interactions with specific content types, a strong user demand for that type is identified. Machine learning algorithms (such as K-means clustering and Apriori algorithm) are applied to model user needs. The model performance is evaluated using cross-validation, and model parameters are adjusted to improve the accuracy of need mining. The mined deep user interaction needs are recorded in a database, ensuring that each user's needs have a corresponding description and label. The needs characteristics of different user groups are analyzed to identify common and individual needs. Design a structure for personalized user profiles, including basic user information, interactive interests and preferences, and deeper needs. Ensure the profile comprehensively reflects user needs and behavioral characteristics. Establish an adaptive profile evolution mechanism to dynamically update based on the latest user interactions and changes in needs. Ensure the user profile always reflects the user's latest state. Set the profile update frequency, such as updating the user profile weekly, or triggering an update when user behavior changes significantly. Ensure the profile responds promptly to changes in user needs. Record the mined personalized user profiles in a database, ensuring consistent data format for easy retrieval and analysis. Each user's profile should include: user ID, interests and preferences, deeper needs, and sentiment indicators. Optimize user profiles by combining user feedback and behavioral data. If users provide positive feedback on a certain type of content, increase the weight of that type of content in the profile.

[0043] Step S4: Perform visual element analysis and multimodal visual mining based on the user's multi-platform interaction content data pool to obtain multimodal visual element information for each homepage;

[0044] In this embodiment, the goal of the visual element parsing model is clearly defined: to extract visual element information from each homepage from a user multi-platform interaction content data pool. This includes data across multiple dimensions such as color, pattern, font, and layout, for comprehensive visual analysis. Homepage data from various platforms is extracted from the data pool, including HTML structure, CSS styles, and image resources. Data integrity is ensured for accurate subsequent visual element parsing. The types of visual elements to be parsed are defined, such as color, shape, font, and layout. Based on the goal, a parsing algorithm for each element is determined, such as using K-means clustering for color extraction and OCR technology for text extraction. Color extraction algorithms are used to analyze the visual elements of the homepage. The dominant color tone of each element is extracted using image processing techniques (such as OpenCV), and the color distribution ratio on the page is calculated. If the homepage is mainly blue and white, these colors and their proportions are recorded. Patterns and shapes on the homepage are analyzed; edge detection and morphological operations are used to identify the shapes of different elements, which helps in recognizing the features of visual elements such as buttons and background patterns. OCR (Optical Character Recognition) technology is used to extract text information from the homepage, and font style, size, and color are recorded. Simultaneously, the layout structure of elements is analyzed to identify their relative positions and arrangements on the page. The extracted visual element information is recorded in a database, ensuring each element has a corresponding label and feature description. The distribution of visual elements across different homepages is analyzed for subsequent comparison and evaluation. The extracted color, pattern, font, and layout information are integrated to form multimodal visual element information for each homepage, comprehensively reflecting its visual style and design characteristics. A multimodal visual information analysis model is constructed, and deep learning techniques (such as convolutional neural networks) are used to train the integrated data. This model should be able to identify relationships between different visual elements and extract key features. The ratio of training to test sets is set (e.g., 80% training, 20% test), and appropriate feature dimensions (e.g., RGB values ​​of colors, font styles, relative positions of layouts) are selected for model training and evaluation. The integrated multimodal visual element information is recorded in a database, ensuring consistent data format for easy subsequent analysis and retrieval. Each homepage should include detailed information on its visual elements. Visualization tools are used to generate charts displaying the multimodal visual elements, helping to intuitively understand the visual characteristics of each homepage. It can generate color distribution maps, layout structure diagrams, and more to showcase different homepage design styles. It allows you to set display dimensions for visualization charts, such as color, font, and layout, to ensure the comprehensiveness and effectiveness of the information presented.

[0045] Step S5: Perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend;

[0046] In this embodiment, key dimensions of user aesthetic and cultural characteristics are defined, including cultural background, aesthetic preferences, and emotional responses. By analyzing users' multimodal visual element information, cultural characteristics related to user aesthetics are identified. Relevant data, including color preferences, design styles, and pattern usage frequency, are extracted from users' visual element information; this data will provide a foundation for subsequent aesthetic and cultural analysis. Classification criteria for aesthetic and cultural characteristics are set; for example, users' color preferences are divided into "warm colors," "cool colors," and "neutral colors," and the aesthetic differences among users under different cultural backgrounds are analyzed. Multimodal visual element information is integrated with users' historical interaction behavior data to analyze users' aesthetic preferences under different cultural dimensions. If users frequently interact with elements containing traditional patterns, it can be inferred that they have a strong traditional cultural aesthetic inclination. Cluster analysis is used to classify users according to their aesthetic and cultural characteristics. K-means or hierarchical clustering algorithms are used to identify user groups with different aesthetic and cultural types, and their characteristics are analyzed. The analysis results are recorded in a database to ensure that each user's aesthetic and cultural characteristics have a corresponding description. Simultaneously, visual charts are used to display the distribution of cultural characteristics of different user groups for easy identification and comparison. Based on users' aesthetic and cultural characteristics, a dynamic aesthetic trend model is constructed. The model should be able to adapt to changes in users' aesthetic preferences over different time periods, reflecting dynamic aesthetic trends. Regularly collecting user interaction data and extracting the latest visual element preferences and aesthetic cultural characteristics will provide real-time data support for fitting dynamic aesthetic trends. Setting a model update frequency, such as monthly, ensures timely reflection of user aesthetic changes. Simultaneously, selecting appropriate feature dimensions for model training, such as user behavior frequency and interactive sentiment, is crucial. Historical data should be used to train the dynamic aesthetic trend model, and cross-validation should be used to evaluate its predictive performance. Model parameters should be optimized to improve fitting accuracy. The fitted dynamic aesthetic trends should be recorded in a database, and a feedback mechanism should be established to adjust the model based on new user behaviors and feedback. If users recently show a stronger preference for a certain visual style, the model will automatically adjust the relevant weights.

[0047] Step S6: Based on the user's dynamic aesthetic trends, perform multi-visual element-driven learning on the personalized user demand profile, and perform feedback compensation optimization to build an intelligent visual element generation model.

[0048] In this embodiment, the user's dynamic aesthetic status information is integrated with personalized user demand profiles. This process involves unifying the user's aesthetic changes, interaction behaviors, and demand characteristics into a comprehensive data framework for subsequent analysis and model building. Key features are extracted from the integrated data, including the user's color preferences, design styles, and emotional responses. These features become the input variables for the intelligent visual element generation model, helping the model understand user needs. Dimensions for feature extraction are set, such as color type, style, and shape. Principal Component Analysis (PCA) and other methods can be used to reduce dimensionality, ensuring that the model's input data is both rich and concise, improving computational efficiency. Based on user demand characteristics, a multi-visual element-driven learning model is designed. Suitable machine learning algorithms, such as Convolutional Neural Networks (CNNs) or Generative Adversarial Networks (GANs) in deep learning, are selected to generate visual elements that conform to the user's aesthetics. The model is trained to generate new visual elements using the user's historical interaction data and feedback information. The training data can be augmented, for example, by rotating or scaling, to increase the diversity of samples and improve the model's generalization ability. Set the ratio of training to test sets (e.g., 80% training, 20% test) and the model's hyperparameters, such as learning rate and batch size. Optimize hyperparameters using methods like grid search to ensure optimal model performance. During user interaction, collect real-time user feedback on generated visual elements. This can be done through user ratings, click-through rates, comments, etc., forming user preference data for visual elements. Design a feedback compensation mechanism to integrate user feedback into the intelligent visual element generation model. Adjust the model's generation strategy based on user feedback, such as increasing the frequency of high-rated elements and decreasing the weight of low-rated elements. Set a feedback update frequency, such as updating model parameters weekly, to ensure the model can respond promptly to changes in user needs. Simultaneously, establish specific rules for feedback compensation, such as setting a threshold; only feedback exceeding this threshold will affect model adjustments. Train the intelligent visual element generation model using the integrated user data and feedback. Through iterative training, continuously optimize model parameters to improve the quality of generated elements and user satisfaction. After model training is complete, evaluate the model using a test set. Multiple metrics, such as the diversity of generated elements and the average user ratings, can be used to comprehensively evaluate the model's generative capabilities. The model's training results and user feedback are recorded, and visualization tools are used to display the distribution of generated elements and changes in user satisfaction. Radar charts are generated to show the ratings of different visual elements, helping to understand changes in user preferences.

[0049] In this embodiment, see Figure 2 The diagram below illustrates the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0050] Acquire multi-dimensional user internet behavior data; perform anomaly cleaning on the internet behavior data to obtain cleaned and optimized internet behavior data;

[0051] Based on the cleaned and optimized internet behavior data, multi-platform content characteristic analysis is performed to obtain the content characteristics of each platform;

[0052] Multi-platform difference identification is performed on the content characteristics to generate difference features for multiple platforms;

[0053] Based on the aforementioned differences, platform types are classified, and content tags are generated for each platform;

[0054] Based on the content tags of each platform, the cleaned and optimized Internet behavior data is mined platform by platform to obtain a multi-platform user interaction content data pool.

[0055] In this embodiment, the target platforms for data collection are defined, covering social media, e-commerce, news websites, video platforms, etc. Each platform should possess rich user behavior data, including clicks, browsing duration, sharing, comments, etc. After obtaining authorization from users and relevant platforms, user behavior data on social media (such as Facebook, Instagram) and shopping records on e-commerce platforms (such as Amazon) are collected. Data crawlers or API interfaces are used to extract user behavior data from each platform. The collected data ensures that it includes user ID, behavior timestamp, behavior type, interaction content, etc. Each platform's crawler program is configured to automatically capture the latest user behavior data every hour and store it in a unified database. After data collection, user behavior data from different platforms is integrated into a unified data framework for easier subsequent processing. The data format is ensured to be consistent for easy analysis. The integrated data framework should include fields such as user ID, platform, behavior time, behavior type, and content ID. Criteria for identifying abnormal data are determined, including data integrity, reasonableness, and consistency. Data with unreasonable timestamps (e.g., future timestamps) or mismatched user behavior types (e.g., purchase behavior without a product ID) is considered abnormal data. If a user has activity records from multiple different platforms within the same timestamp, an anomaly may exist. Data cleaning algorithms are employed to identify and remove outlier data. Rule-based filtering and statistical analysis (such as Z-score) can be used to identify outliers. If a user is found to have more than 100 activity records in a single day, and the activity type is primarily "browsing," these records are marked as anomalies. The cleaned data results are recorded to ensure the integrity and accuracy of the dataset. The cleaned results are verified through sampling to ensure no residual outlier data. Changes in data volume before and after cleansing are recorded. If the original data set contains 10,000 records and 8,500 records remain after cleansing, this change is recorded, and the reasons for removal are analyzed.

[0056] Define the dimensions of content characteristic analysis, including content type (video, text, image), theme (technology, fashion, entertainment), and interactivity (number of comments, number of shares). Set the content type to "text and images" or "video," and the theme to "news" or "entertainment." Process the cleaned and optimized internet behavior data using text analysis and data mining techniques to extract the content characteristics of each platform. Natural Language Processing (NLP) techniques can be used for text analysis. For social media platforms, extract keywords from user comments, identify high-frequency words and themes to understand the main content of user interactions. Record the analyzed content characteristics in a database and use visualization tools to display the distribution of content characteristics across different platforms to help understand the content characteristics of different platforms. Generate radar charts to show the performance of each platform in multiple dimensions (such as interactivity and content type) for easy comparison. Select appropriate statistical analysis methods (such as ANOVA, t-test, etc.) to identify differences in content characteristics across different platforms and determine significant differences in content characteristics. Compare the differences in user interaction methods between social media and e-commerce platforms to see if there are significant differences in the number of comments, browsing time, etc.

[0057] Based on the statistical analysis results, differentiated features of multiple platforms are generated. These features will help identify the unique characteristics of each platform, thus providing a basis for subsequent content tag generation. If the analysis results show that social media platforms are mainly based on images and short videos, while e-commerce platforms are mainly based on user reviews and purchasing behavior, these differentiated features are recorded. These differentiated features are recorded in the system and displayed using visualization tools to facilitate understanding of the differences between platforms. Bar charts are generated to show the differences between different platforms in terms of content type and interactivity, helping to identify the characteristics of each platform. Based on the differentiated features from the previous step, platform type classification criteria are set. Classification can be based on content characteristics, user interaction behavior, etc., for example, platforms can be divided into social, transactional, and informational types. Social platforms are defined as platforms with strong content interactivity, and e-commerce platforms as platforms primarily focused on transactions. Each platform is classified, and content tags for each platform are generated using machine learning algorithms (such as K-means clustering) or rule-based classification methods. If a platform's content characteristics show strong interactivity and a focus on text and images, it is labeled as a "social interaction platform." Platform categories and content tags are recorded in the database and verified to ensure classification accuracy. Cross-validation was performed to ensure the consistency and reliability of the classification results. User interaction behaviors were clearly defined, including likes, comments, shares, and browsing. Based on content tags, user interaction behaviors across different platforms were identified. Social media platform interactions were defined as including comments and sharing, while e-commerce platform interactions included purchases and reviews. User interaction behavior was mined platform-by-platform based on cleaned and optimized internet behavior data and platform content tags. Data mining techniques (such as association rule analysis) were used to identify user behavior patterns. Analyzing user behavior on social media platforms revealed that users interacted more frequently with specific topics within a certain timeframe. The mined user interaction content data from multiple platforms was integrated into a user interaction behavior pool to provide data support for subsequent personalized recommendations. Each user's interaction behavior across different platforms was recorded to form a user interaction data pool, facilitating the analysis and mining of user preferences.

[0058] In this embodiment, see Figure 3 The diagram below illustrates the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0059] Key interactive fonts were extracted from the user's multi-platform interaction content data pool to obtain multiple key interactive content fonts;

[0060] Perform deep semantic meaning analysis on the fonts of the key interactive content to generate deep semantic features for each font;

[0061] Based on the deep semantic features, user interaction intent is inferred and generated.

[0062] Personalized interaction emotion recognition is performed on user interaction intent to generate personalized interaction emotions for users;

[0063] Personalized interaction interest preferences are mined based on a user's multi-platform interaction content data pool to obtain user interaction interest preferences.

[0064] In this embodiment, user interaction content, including comments, shares, and likes, is extracted from a multi-platform user interaction content data pool. Data integrity and consistency are ensured by removing duplicate records and outliers for subsequent analysis. If the data pool contains records of a user's comments and likes on specific posts on social media, these records must be accurate and unique. Text-based algorithms (such as TF-IDF and TextRank) are used to extract key interactive fonts from the user interaction content. TF-IDF measures the importance of words in a document, thus identifying the most frequently mentioned keywords. Keywords in user comments are analyzed; if "environmental protection" appears frequently in multiple comments, it is identified as a key interactive font. The extracted key interactive fonts are recorded in a database and visualized to help intuitively understand user interaction focus. Suitable Natural Language Processing (NLP) tools are selected for deep semantic analysis. Models such as Word2Vec and BERT can be used to capture the contextual relationships and semantic information of words. The BERT model is used to analyze the word "environmental protection," identifying its meaning in different contexts, such as "environmentally friendly products" and "environmentally friendly policies."

[0065] Deep semantic analysis is performed on the extracted key interactive fonts to generate deep semantic features for each font. Word vector representation and contextual relevance are considered to extract semantic similarity and related topics. The model calculates a similarity of 0.85 between "environmental protection" and "sustainable development," indicating a strong correlation between the two in user interactions. Based on the deep semantic features, a user interaction intent inference model is constructed. Classification models (such as decision trees and support vector machines) can be used to classify user interaction behaviors and infer their potential intents. User comments are matched with known intents (such as "purchase intent" and "information acquisition intent") to train the model and improve accuracy. The inference model analyzes user interaction content to identify each user's interaction intent. If a user frequently comments on content related to "environmental protection," their intent is inferred to be an interest in environmentally friendly products. If a user discusses "how to reduce plastic use" on multiple platforms, their intent is inferred to be "advocating for environmental protection." The inferred user interaction intents are recorded in a database and validated to ensure the accuracy of intent inference. Cross-validation is used to improve the model's reliability. The record format is: User ID: 001, Interaction Intent: "Focus on Environmentally Friendly Products", Validation Result: High Accuracy. A suitable sentiment analysis model (such as a sentiment dictionary approach or a deep learning model) is selected to identify the user's personalized interactive sentiment. Sentiment analysis will help understand the user's emotional response to specific topics. An LSTM model is used to analyze the sentiment tendency of user comments, identifying whether the user's sentiment is positive, negative, or neutral.

[0066] Sentiment analysis is used to extract personalized emotional characteristics of users during interactions. Analyzing user comments on the topic of "environmental protection" identifies their emotional inclination as positive, negative, or neutral. For example, a user commenting "I like this environmentally friendly product, it's great" indicates a positive sentiment. Personalized interaction sentiment characteristics are recorded in the system and visualized in charts to show the distribution of user sentiment across different topics. The definition of personalized interaction interest preferences is clarified, including user preferences for different topics, content types, and interaction methods. Interest preferences can be identified by analyzing user interaction content. If a user frequently interacts with content related to "technological innovation," their interest preference can be defined as "technology." Based on user interaction behavior and emotional characteristics, personalized user interest preferences are mined. Cluster analysis and other methods can be used to segment users into different interest groups. K-means clustering is used to categorize user interests into different categories such as "environmental protection," "technology," and "fashion." The mined user interaction interest preferences are recorded in a database, and a feedback mechanism is established to ensure dynamic adjustment of recommendation strategies in subsequent recommendations. The record format is: User ID: 001, Interest Preference: "Environmental Protection," Status: Highly Relevant, facilitating subsequent personalized recommendations.

[0067] In this embodiment, the specific steps for mining personalized interaction interest preferences based on the user's multi-platform interaction content data pool to obtain the user's interaction interest preferences are as follows:

[0068] The interaction frequency and click volume of each platform are calculated based on a user multi-platform interaction content data pool.

[0069] Extract the timestamp of each interaction behavior from a user's multi-platform interaction content data pool;

[0070] Based on the timestamps, user interaction time distribution analysis is performed to obtain interaction time distribution characteristics;

[0071] Based on the interaction frequency, click volume and interaction time distribution characteristics, multi-dimensional interaction behavior feature analysis is performed to generate user multi-dimensional interaction behavior patterns.

[0072] By mining users' multi-dimensional interaction behavior patterns to identify their personalized interaction interests and preferences, we can obtain users' interaction interest preferences.

[0073] In this embodiment, user behavior data for each platform, including interactions such as comments, likes, and shares, is extracted from a multi-platform user interaction content data pool. Data integrity is ensured by removing duplicate records to improve accuracy. Assuming the collected data includes user interactions on social media, e-commerce, and video platforms, the record format should include fields such as user ID, platform, behavior type, and timestamp. Interaction behavior on each platform is statistically analyzed, calculating the total interaction frequency and total clicks for each platform. Interaction frequency refers to the number of times a user performs an action on that platform, while clicks include the number of links or content clicked by the user. If a user on a social media platform performs 200 interactions (likes, comments, etc.) and clicks 150 links in the past week, the social media interaction frequency is recorded as 200, and the click count as 150. The calculated interaction frequency and click count are recorded in the database, and a visualization chart is generated to help analyze user interactions across different platforms. A bar chart is generated, with the platform name on the X-axis and interaction frequency and click count on the Y-axis, clearly displaying user activity on each platform. Timestamp Extraction: This step extracts the timestamp of each interaction from the user's multi-platform interaction content data pool. This step obtains the time information of user interactions to facilitate subsequent analysis of the time distribution of these interactions. For example, assuming a user commented on social media with a timestamp of 2023-04-01 08:00:00, record the timestamp for each interaction. Organize the extracted timestamps into a unified format to ensure data timeliness and consistency. Standard time formats (such as ISO 8601) can be used to store timestamps. The timestamp format should be standardized to "YYYY-MM-DD HH:MM:SS" to ensure all recorded time information can be directly analyzed. Record the extracted timestamps in a database and visualize them to help analyze the time characteristics of user interactions. Based on the extracted timestamp data, statistically analyze the distribution of user interactions across different time periods. Time can be divided into hours, days, or weeks to observe changes in interaction behavior. Divide a day into 24 hours and count the number of user interactions each hour to form a time distribution dataset. Analyze the time distribution data to extract interaction time distribution characteristics, including peak periods, active periods, and inactive periods. It can calculate the proportion of interactive behaviors in each time period to identify the time periods when users are most active. If it is found that the frequency of user interactions is significantly higher between 8:00 and 9:00 AM than at other times, then this time period is recorded as the peak period.

[0074] The system records the distribution characteristics of user interaction time and visualizes them through charts to help intuitively understand user activity levels at different times. A heatmap is generated, with the X-axis representing time periods and the Y-axis representing interaction frequency. Color intensity indicates interaction strength, clearly revealing the time patterns of user interaction. Indicators for multi-dimensional interactive behavior characteristics are determined, including interaction frequency, click volume, time distribution, and content type. These indicators are combined to form a multi-dimensional user interaction profile. Interaction frequency is defined as the number of times a user interacts on a specific platform, and click volume is defined as the number of pieces of content a user clicks. Based on interaction frequency, click volume, and time distribution characteristics, in-depth analysis of user interaction behavior is conducted. Cluster analysis or correlation analysis can be used to identify user behavior patterns and trends. If a user frequently interacts on a social platform and is active between 8 PM and 10 PM, their characteristic is recorded as an "evening active user." Based on multi-dimensional interactive behavior characteristics, a personalized user interaction interest preference model is constructed. Machine learning algorithms (such as collaborative filtering and content recommendation) can be used to identify interest preferences. Combining user interaction frequency and content type, the model is trained to identify user interests on specific topics. The model is used to analyze user interaction behavior and uncover personalized user interests and preferences. Identify user preferences across different platforms and content types. If the model analyzes that a user interacts more frequently with "technology"-related content than with other topics, record the user's interest preference in "technology." Record the mined user interaction interest preferences in the system and establish a feedback mechanism to ensure that recommendation strategies can be dynamically adjusted in subsequent recommendations. The recording format is: User ID: 001, Interest Preference: "Technology", Status: Highly Relevant, facilitating subsequent personalized recommendations.

[0075] In this embodiment, see Figure 4 The diagram below illustrates the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0076] Quantitative analysis of interaction topology correlation was conducted based on user personalized interaction sentiment and user interaction interest preferences to obtain the quantitative relationship between interaction sentiment and interest preferences.

[0077] Based on the quantitative relationship, in-depth interaction needs are mined to generate deep user interaction needs.

[0078] Based on the content tags of each platform, potential multi-scenario demand predictions are made for users’ deep interaction needs, and multi-scenario demand prediction features are generated.

[0079] Adaptive profiling evolution is performed on multi-scenario demand prediction features to construct personalized user demand profiles.

[0080] In this embodiment, user's personalized interaction sentiment and interaction interest preference data are integrated into a unified data framework. The data ensures it includes user ID, interaction sentiment rating (e.g., positive, negative, neutral), and interaction interest preferences (e.g., technology, entertainment, health). A data table is constructed, containing fields: user ID, sentiment rating, and interest preference. Assume user 001 has a sentiment rating of 0.8 (positive) and an interest preference of "technology". Statistical analysis methods (e.g., correlation analysis, regression analysis) are used to calculate the quantitative relationship between interaction sentiment and interest preference. By calculating the statistical correlation between sentiment rating and interest preference, the degree of association between the user's emotional attitude and their interests can be identified. If it is found that the average positive sentiment rating for "technology" related content is 0.75, while the negative sentiment rating is 0.2, it can be concluded that the user has strong positive sentiment in this area. Based on the quantitative relationship between interaction sentiment and interest preference, a deep interaction demand mining model is constructed. This model can use cluster analysis or association rule analysis to identify the user's potential needs. The Apriori algorithm is used to analyze user interaction behavior, identify frequently occurring sentiment-interest combinations, and thus infer deep-seated needs. Through model analysis, we identify users' deep-seated interaction needs. These needs may not be directly reflected in their interaction behavior, but can be inferred through a combination of emotions and interests. If a user has a high emotional score in the "technology" field and frequently interacts with related content, we can infer their demand for new technology products. We record these deep-seated user interaction needs in a database and verify their accuracy through user feedback. We can conduct small-scale tests to observe user reactions to recommended content. Based on the content tags of each platform, we identify potential multi-scenario needs related to users' deep-seated interaction needs. Content tags should cover the platform's theme, content type, and user interaction methods. We label "technology" related content as "latest technology news," "smart home products," etc., ensuring that content tags match user needs. We build a demand prediction model to predict users' potential needs in different scenarios based on their deep-seated interaction needs and platform content tags. Machine learning algorithms (such as random forests and support vector machines) can be used for demand prediction. If the model detects a user's deep-seated need for "smart home," it can predict that the user may develop an interest in related products in the future. The system records multi-scenario demand prediction characteristics and generates visual charts to help analyze user demand prediction results in different scenarios. It generates trend charts to show changes in user demand across different scenarios (such as home, office, and entertainment), facilitating understanding of users' potential interests. The system defines the components of a personalized user profile, including basic user information, interaction behavior characteristics, interests, and underlying needs. It ensures the profile comprehensively reflects user needs and behavioral characteristics. The user profile definition includes: user ID, interests (such as technology and health), and underlying needs (such as smart home products).Design an adaptive evolution mechanism for user profiles, enabling them to dynamically update based on users' latest interactions and changing needs. This ensures user profiles always reflect the latest user status. If a user's interaction frequency with "smart home" increases recently, their underlying needs are automatically updated to "interest in smart home products." The evolved personalized user need profile is recorded in the system and applied to the personalized recommendation system to optimize recommendation effectiveness. User ID: 001, Personalized Need Profile: {Interest Preference: "Technology", Underlying Need: "Smart Home Products"}, ensuring that recommended content is dynamically adjusted based on this profile during recommendation.

[0081] In this embodiment, step S4 includes the following steps:

[0082] Based on a user multi-platform interaction content data pool, locate homepages on multiple platforms;

[0083] Visual element recognition is performed on the homepages of the multiple platforms, and the visual communication elements of each homepage are marked;

[0084] Visual element analysis is performed based on the visual communication elements to extract the color, pattern, font, and layout information of the homepage visual elements;

[0085] The colors and patterns are analyzed to generate a homepage visual color scheme.

[0086] Based on the font and layout information, element morphology feature analysis is performed to generate homepage element morphology characteristics;

[0087] Multimodal visual mining is performed on the homepage's visual color scheme and element shape characteristics to obtain multimodal visual element information for each homepage.

[0088] In this embodiment, multiple platforms to be analyzed are identified, such as social media, e-commerce, and news websites. Access to the homepages of these platforms is ensured for subsequent visual element identification and analysis. Target platforms may include Facebook, Instagram, Amazon, and CNN, and the homepage URL for each platform is recorded. Web scraping tools are configured to automate access to and acquisition of the homepage content of the selected platforms. This can be achieved using web scraping techniques or API interfaces, ensuring the complete homepage HTML structure can be extracted. Python libraries such as Scrapy or Beautiful Soup are used to scrape the homepage HTML and extract the required elements. The scraped homepage data is organized and stored in a unified database. The data format is ensured to be consistent for subsequent analysis. The homepage HTML, CSS, and related metadata for each platform are recorded. The recording format includes: platform name, homepage URL, scraping time, etc., for traceability during subsequent analysis. Appropriate visual element recognition tools, such as computer vision libraries (OpenCV, Tesseract, etc.), are selected to extract visual elements from the homepage. This tool should be able to recognize information such as images, colors, fonts, and layout. OpenCV is used for image processing to identify visual elements such as buttons, images, and backgrounds on the homepage. The identified visual elements are labeled and categorized, including different types such as images, buttons, titles, and text boxes. Each element is ensured to have a corresponding label for subsequent analysis. If a button is detected, its type is recorded as "button," its location as "top left corner," and its color and text are labeled. The labeled visual elements are recorded in a database, and visualizations are generated to show the distribution of visual elements on each homepage, helping to understand the visual design styles of different platforms. Mind maps or flowcharts are generated to show the visual element structure of each platform, facilitating the analysis of its design characteristics. Color extraction algorithms (such as K-means clustering) are used to extract the primary colors from the visual elements, and the proportion of each color on the homepage is calculated, which helps identify the main color tone and color scheme of the homepage. The homepage background color, button color, and text color are analyzed, and clustering algorithms identify blue as the primary color and white as the secondary color. Patterns (such as textures and background patterns) and layout information (such as element arrangement and alignment) on the homepage are analyzed. Image processing techniques can be used to identify repeating patterns and layout structures. If the homepage uses a grid layout, the position and size of each element within the grid will be recorded, and the symmetry and uniformity of the layout will be analyzed. The parsed color, pattern, and layout information will be recorded in a database, and visual charts will be generated to display the visual element characteristics of the homepage. A color distribution map will be generated to show the proportion of different colors on the homepage, and a layout diagram will be drawn to show the spatial distribution of elements. Based on the extracted color information, an overall color tone analysis will be performed to identify the visual color tone style of the homepage.Color theory (such as color wheels and color schemes) can be used for classification, such as warm, cool, or neutral tones. If the homepage uses blue and green as its main colors, it can be classified as a "cool" style. The identified visual color styles are categorized, and a description is generated for each style for subsequent application in personalized recommendation systems. The record format is "Homepage ID:001, Color Style: Cool, Description: Mainly blue and green, giving a fresh and tranquil feeling." Analytical indicators for element morphological characteristics are determined, including the element's shape (e.g., circle, square), corners (e.g., rounded corners, right angles), and aspect ratio (e.g., width-to-height ratio). If a button is analyzed as a rounded rectangle, its dimensions are recorded as 100px wide and 50px high. Based on the extracted layout information, the morphological features of the homepage elements are analyzed, and the morphological characteristics of each element are recorded and classified. If multiple buttons are identified as circular and evenly distributed in the layout, they are recorded as having the "circular button" feature. The morphological feature analysis results are recorded in a database, and visual charts are generated to help understand the element morphological characteristics of different homepages. Generate bar charts to display the number of elements with different morphological characteristics, intuitively reflecting the design style of each platform. Integrate the visual color scheme and element morphological characteristics of the homepage to construct multimodal visual element information. Ensure the data includes information across multiple dimensions such as color scheme, shape, and layout. The record format is "Homepage ID: 001, Color Scheme: Cool Color Scheme, Element Shape: Circular Button, Layout: Grid". Analyze the integrated multimodal visual element information to identify the design features of the homepage in different dimensions. Cluster analysis and other methods can be used to identify homepages with similar design styles. Analysis reveals that some homepages share commonalities in the use of cool colors and circular elements, allowing them to be grouped together. Record the multimodal visual element information in the database and display the comprehensive visual characteristics of different homepages through visualization charts. Generate radar charts to comprehensively display the performance of each homepage in multiple dimensions such as color scheme, shape, and layout, facilitating subsequent personalized recommendations.

[0089] In this embodiment, step S5 includes the following steps:

[0090] Based on the user's personalized interactive emotions, positive and negative elements are inferred from the multimodal visual element information to identify the positive and negative value of each visual element information;

[0091] Based on the aforementioned positive and negative values, user aesthetic culture analysis is conducted to generate aesthetic culture characteristics of positive and negative elements;

[0092] Based on the aesthetic and cultural characteristics of positive and negative elements, a differential preference analysis of positive and negative elements is conducted to obtain the differential preference characteristics of positive and negative elements;

[0093] Dynamic aesthetic trends are fitted based on the differentiated preference characteristics of positive and negative elements to construct dynamic aesthetic trends of users.

[0094] In this embodiment, sentiment data is extracted from user interaction behavior, recording user feedback for each visual element, including likes, comments, and shares. The integrity and accuracy of the data are ensured for subsequent analysis. If a user comments on a visual element as "I really like this design," its sentiment score is recorded as positive; if the comment is "I don't like this color," it is recorded as negative. Sentiment analysis models (such as sentiment lexicon or deep learning models) are used to analyze user feedback and infer the positive or negative value of each visual element. This model should be able to identify and quantify the sentiment tendency in the text. The BERT model is used to analyze user comments; if the model's output sentiment score is higher than 0.5, it is marked as a positive element, and lower than 0.5, it is marked as a negative element. Based on the inferred positive and negative elements, analytical indicators of aesthetic cultural characteristics are defined, including cultural background, aesthetic preferences, and emotional responses. These characteristics will help identify the user's aesthetic culture. Cultural characteristic dimensions are set as: traditional culture, modern design, color preferences, etc. Statistical analysis of the positive and negative values ​​of each visual element is performed to identify user preferences for different cultural characteristics. Cluster analysis methods can be used to divide users into different aesthetic culture groups. If a user is found to favor positive elements related to traditional culture, their aesthetic cultural characteristic is recorded as "traditional aesthetics." This aesthetic cultural characteristic is recorded in a database, and its distribution across different users is displayed through charts to understand the aesthetic preferences of the user group. A radar chart is generated to show different users' preferences for traditional and modern designs, clearly presenting their aesthetic cultural characteristics. A method for differentiated preference analysis is designed to identify differences in user preferences for positive and negative elements. Statistical methods such as t-tests or ANOVA can be used for comparative analysis. User ratings for positive and negative elements in terms of color, shape, and layout are compared to analyze the differences. Based on the analysis results, differentiated preference features for positive and negative elements are extracted, including visual characteristics and design styles, which helps identify subtle differences in user aesthetics. If a user prefers bright colors for positive elements and dark colors for negative elements, this differentiated feature is recorded. This differentiated preference feature is recorded in a database, and the analysis results are displayed through visual charts to help understand the differences in user preferences for positive and negative elements. A double bar chart is generated to compare the ratings of positive and negative elements across various feature dimensions, visually presenting the differences in user preferences. Based on the differentiated preference characteristics of positive and negative elements, a dynamic aesthetic trend model for users is constructed. This model should be able to reflect changes in users' aesthetic preferences in real time, adapting to user feedback and needs. Time series analysis is used to fit users' aesthetic trends to identify their changing aesthetic tendencies over different time periods. The dynamic aesthetic trend model is trained and its predictive ability optimized through user interaction data and aesthetic feedback. The model should be able to predict users' future aesthetic preferences based on their historical behavior.If user sentiment feedback over the past month shows an increased preference for vibrant colors, the model will adjust its future color recommendation strategy. The fitted dynamic aesthetic trends are recorded in a database, and the model's effectiveness is validated by assessing its predictive accuracy. A / B testing and other methods can be used to observe the model's performance. User satisfaction ratings after using the new model are recorded; a significant improvement confirms the model's effectiveness.

[0095] In this embodiment, step S6 includes the following steps:

[0096] Based on the dynamic aesthetic trends of users, we conduct multi-visual element-driven learning of personalized user needs profiles to build a dynamic visual element recommendation engine.

[0097] Multiple visual communication simulation elements are generated based on the dynamic visual element recommendation engine and displayed to users in a visual manner, while collecting real-time user feedback.

[0098] Analyze real-time user feedback to understand user feedback requirements;

[0099] Based on user feedback, the dynamic visual element recommendation engine is optimized through feedback compensation to build an intelligent visual element generation model.

[0100] In this embodiment, historical user interaction data, emotional feedback, and preference information are collected to analyze users' dynamic aesthetic trends. This data may include user behavior records, interactive emotions, and interest preferences across different platforms. User activity over the past month is analyzed, recording the frequency of interaction and emotional ratings for different visual elements (such as color, shape, and layout) to ensure data integrity and accuracy. Based on the collected dynamic aesthetic trend data, a multi-visual element-driven learning model is constructed. Deep learning algorithms (such as convolutional neural networks) can be used to identify and learn users' visual preferences. The model is trained using user interaction data to identify user preference patterns on visual elements, such as a preference for warm colors and minimalist layouts. The trained model is integrated into a recommendation engine to implement dynamic visual element recommendation functionality. This engine should be able to analyze users' aesthetic trends in real time and adjust recommendations accordingly. If the model identifies an increase in users' recent preference for "natural style" elements, the recommendation engine will prioritize recommending related visual elements. A visual element generation mechanism is designed to automatically generate multiple visual communication simulation elements based on the output of the recommendation engine. These elements should cover different styles, tones, and shapes to meet users' diverse needs. The generated simulated elements can include buttons, background patterns, font styles, etc., with different color combinations to provide multiple choices. These visual communication simulated elements are then presented to users through a visual interface. A simple and intuitive user interface is designed to allow users to easily browse and select different visual elements. A visual display page is created containing thumbnails of various simulated elements, which users can click to view detailed information. User feedback is collected in real time as they browse and select visual elements; this can be done through questionnaires, rating systems, or click behavior. After viewing simulated elements, users can be asked to rate their satisfaction (e.g., 1 to 5 points), and their choices and preferences are recorded. The collected real-time user feedback is organized and categorized. Data format consistency is ensured for easy subsequent analysis. Each user's rating, selection, and feedback on different simulated elements are recorded. A data table is created, containing fields such as user ID, element ID, rating, and comment, ensuring data integrity. Based on user feedback, a feedback demand analysis model is constructed. Natural language processing techniques can be used to analyze user text comments, extracting keywords and sentiment tendencies. If a user comments "This color looks good," it is analyzed as positive feedback, and the color preference is recorded. The parsed user feedback is recorded in a database and visualized using charts to help understand overall user feedback trends. Bar charts are generated to display the average ratings of different visual elements and the sentiment distribution of user feedback, helping to optimize recommendation strategies. Based on user feedback, the dynamic visual element recommendation engine is optimized through feedback compensation. A mechanism is designed to allow the engine to dynamically adjust based on negative user feedback. If a user rates a certain element low, its frequency of appearance is reduced in subsequent recommendations.Based on a feedback compensation mechanism, an intelligent visual element generation model is constructed. This model should be able to adapt to changes in user needs and dynamically adjust the style and type of generated visual elements. If users have shown a preference for "minimalist design" in past feedback, the model will prioritize generating visual elements that conform to this style. The effects of the optimized recommendation engine and generation model are recorded in the system and evaluated. The impact of the new model on user satisfaction and interaction behavior is observed through methods such as A / B testing. Changes in user satisfaction ratings after using the new model are recorded; if the improvement is significant, the optimization is considered successful.

[0101] In this embodiment, a personalized visual communication element recommendation system based on Internet big data is provided, used to execute the personalized visual communication element recommendation method based on Internet big data as described above, including:

[0102] The behavior mining module is used to acquire multi-dimensional internet behavior data of users and to mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool.

[0103] The interest and preference module is used to infer user interaction intent from the user's multi-platform interaction content data pool and to mine personalized interaction interest preferences to obtain user interaction interest preferences.

[0104] The user profile module is used to deeply mine user interaction needs based on user interaction interests and preferences, and to adaptively evolve user profiles to build personalized user needs profiles.

[0105] The visual element parsing module is used to parse visual elements and mine multimodal visuals based on the user's multi-platform interaction content data pool, so as to obtain multimodal visual element information for each homepage.

[0106] The aesthetic trend analysis module is used to perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend.

[0107] The feedback compensation and optimization module is used to learn multi-visual element-driven models of personalized user needs profiles based on the user's dynamic aesthetic trends, and to perform feedback compensation and optimization to build an intelligent visual element generation model.

[0108] This invention monitors user behavior across multiple internet platforms (such as social media, e-commerce websites, and news platforms) to obtain rich user interaction data, including clicks, browsing, comments, and sharing, laying a solid foundation for subsequent analysis. Mining user interactions across different platforms helps identify behavioral patterns and habits, understand their reactions in different contexts, and thus provide a basis for personalized recommendations. By analyzing the user interaction content data pool, it's possible to infer user interests and intentions, providing foundational data for personalized recommendation systems and helping the system more accurately understand user needs. Deeply mining user interaction interests and preferences helps platforms recommend content that better matches user interests, thereby improving user satisfaction and engagement. Through in-depth needs mining, dynamic personalized user profiles can be generated, reflecting users' latest needs and preferences. These profiles evolve with changes in user behavior, maintaining their timeliness and accuracy. The construction of personalized user needs profiles helps businesses develop more targeted strategies for product promotion and marketing, thereby improving conversion rates. Analyzing the visual elements of user interaction content yields rich multimodal visual information, including images, videos, and graphics, providing multi-dimensional data support for understanding user aesthetics and preferences. Multimodal visual mining enables recommendation systems to offer more diverse and richer visual content, enhancing the user experience. By analyzing users' aesthetic and cultural backgrounds, the platform can understand the aesthetic preferences of users from different cultural backgrounds, thus making recommendations more aligned with users' cultural identity. Constructing a dynamic aesthetic profile of users can reflect changes in their aesthetic needs in real time, helping the platform adjust recommendation strategies promptly and enhancing user experience. Based on the dynamic aesthetic profile of users, multi-visual element-driven learning can establish an intelligent visual element generation model. This model can automatically generate visual content that matches user preferences, improving the intelligence level of recommendations. A feedback compensation optimization mechanism can continuously improve recommendation performance based on actual user feedback, ensuring that the system's adaptability remains consistent with changes in user needs.

[0109] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0110] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A personalized visual communication element recommendation method based on Internet big data, characterized in that, Includes the following steps: Step S1: Obtain multi-dimensional internet behavior data of users and mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool; Step S2: Infer user interaction intent from the user's multi-platform interaction content data pool and mine personalized interaction interest preferences to obtain user interaction interest preferences; Step S3: Based on user interaction interests and preferences, conduct in-depth interaction needs mining and adaptive profile evolution to build personalized user needs profiles; Step S4: Perform visual element analysis and multimodal visual mining based on the user's multi-platform interaction content data pool to obtain multimodal visual element information for each homepage; Step S5: Perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend; Step S6: Based on the user's dynamic aesthetic trends, perform multi-visual element-driven learning on the personalized user demand profile, and perform feedback compensation optimization to build an intelligent visual element generation model. The specific steps of step S5 are as follows: Based on the user's personalized interactive emotions, positive and negative elements are inferred from the multimodal visual element information to identify the positive and negative value of each visual element information; Based on the aforementioned positive and negative values, user aesthetic culture analysis is conducted to generate aesthetic culture characteristics of positive and negative elements; Based on the aesthetic and cultural characteristics of positive and negative elements, a differential preference analysis of positive and negative elements is conducted to obtain the differential preference characteristics of positive and negative elements; Dynamic aesthetic trends are fitted based on the differentiated preference characteristics of positive and negative elements to construct dynamic aesthetic trends of users; The specific steps of step S6 are as follows: Based on the dynamic aesthetic trends of users, a dynamic visual element recommendation engine is constructed by learning multi-visual element-driven learning of personalized user needs profiles. Multiple visual communication simulation elements are generated based on the dynamic visual element recommendation engine and displayed to users in a visual manner, while collecting real-time user feedback. Analyze real-time user feedback to understand user feedback requirements; Based on user feedback, the dynamic visual element recommendation engine is optimized through feedback compensation to build an intelligent visual element generation model.

2. The personalized visual communication element recommendation method based on Internet big data according to claim 1, characterized in that, The specific steps of step S1 are as follows: Acquire multi-dimensional user internet behavior data; perform anomaly cleaning on the internet behavior data to obtain cleaned and optimized internet behavior data; Based on the cleaned and optimized internet behavior data, multi-platform content characteristic analysis is performed to obtain the content characteristics of each platform; Multi-platform difference identification is performed on the content characteristics to generate difference features for multiple platforms; Based on the aforementioned differences, platform types are classified, and content tags are generated for each platform; Based on the content tags of each platform, the cleaned and optimized Internet behavior data is mined platform by platform to obtain a multi-platform user interaction content data pool.

3. The personalized visual communication element recommendation method based on Internet big data according to claim 1, characterized in that, The specific steps of step S2 are as follows: Key interactive fonts were extracted from the user's multi-platform interaction content data pool to obtain multiple key interactive content fonts; Perform deep semantic meaning analysis on the fonts of the key interactive content to generate deep semantic features for each font; Based on the deep semantic features, user interaction intent is inferred and generated. Personalized interaction emotion recognition is performed on user interaction intent to generate personalized interaction emotions for users; Personalized interaction interest preferences are mined based on a user's multi-platform interaction content data pool to obtain user interaction interest preferences.

4. The personalized visual communication element recommendation method based on Internet big data according to claim 3, characterized in that, The specific steps for mining personalized interaction interest preferences based on a user multi-platform interaction content data pool to obtain user interaction interest preferences are as follows: The interaction frequency and click volume of each platform are calculated based on a user multi-platform interaction content data pool. Extract the timestamp of each interaction behavior from a user's multi-platform interaction content data pool; Based on the timestamps, user interaction time distribution analysis is performed to obtain interaction time distribution characteristics; Based on the interaction frequency, click volume and interaction time distribution characteristics, multi-dimensional interaction behavior feature analysis is performed to generate user multi-dimensional interaction behavior patterns. By mining users' multi-dimensional interaction behavior patterns to identify their personalized interaction interests and preferences, we can obtain users' interaction interest preferences.

5. The personalized visual communication element recommendation method based on Internet big data according to claim 1, characterized in that, Step S3 is as follows: Quantitative analysis of interaction topology correlation was conducted based on user personalized interaction sentiment and user interaction interest preferences to obtain the quantitative relationship between interaction sentiment and interest preferences. Based on the quantitative relationship, in-depth interaction needs are mined to generate deep user interaction needs. Based on the content tags of each platform, potential multi-scenario demand predictions are made for users’ deep interaction needs, and multi-scenario demand prediction features are generated. Adaptive profiling evolution is performed on multi-scenario demand prediction features to construct personalized user demand profiles.

6. The personalized visual communication element recommendation method based on Internet big data according to claim 1, characterized in that, The specific steps of step S4 are as follows: Based on a user multi-platform interaction content data pool, locate homepages on multiple platforms; Visual element recognition is performed on the homepages of the multiple platforms, and the visual communication elements of each homepage are marked; Visual element analysis is performed based on the visual communication elements to extract the color, pattern, font, and layout information of the homepage visual elements; The colors and patterns are analyzed to generate a homepage visual color scheme. Based on the font and layout information, element morphology feature analysis is performed to generate homepage element morphology characteristics; Multimodal visual mining is performed on the homepage's visual color scheme and element shape characteristics to obtain multimodal visual element information for each homepage.

7. A personalized visual communication element recommendation system based on Internet big data, characterized in that, The method for recommending personalized visual communication elements based on Internet big data as described in claim 1 includes: The behavior mining module is used to acquire multi-dimensional internet behavior data of users and to mine user interaction behavior on each platform to obtain a multi-platform user interaction content data pool. The interest and preference module is used to infer user interaction intent from the user's multi-platform interaction content data pool and to mine personalized interaction interest preferences to obtain user interaction interest preferences. The user profile module is used to deeply mine user interaction needs based on user interaction interests and preferences, and to adaptively evolve user profiles to build personalized user needs profiles. The visual element parsing module is used to parse visual elements and mine multimodal visuals based on the user's multi-platform interaction content data pool, so as to obtain multimodal visual element information for each homepage. The aesthetic trend analysis module is used to perform user aesthetic culture analysis and dynamic aesthetic trend fitting on the multimodal visual element information to construct the user dynamic aesthetic trend. The feedback compensation and optimization module is used to learn multi-visual element-driven models of personalized user needs profiles based on the user's dynamic aesthetic trends, and to perform feedback compensation and optimization to build an intelligent visual element generation model.

Images