Information data processing method and device, and electronic equipment

Abstract

The application relates to the field of data processing, in particular to an information data processing method and device and electronic equipment. The method comprises the following steps: obtaining information data to be analyzed; using a hot topic detection model to identify hot topics in the information data in real time, tracking the change trend of the hot topics in a social network dynamic graph, and obtaining an information analysis result of the hot topics; generating a real-time interaction strategy corresponding to the hot topics based on the information analysis result and the types of various social media platforms; executing the real-time interaction strategy through a real-time interaction model to simulate the behavior characteristics of users in various social media platforms, to obtain matched information to be published and / or behavior operation instructions, and to push the information to the users for confirmation of whether to execute. The method can flexibly adjust the strategy according to different platform characteristics and dynamic information, enhance the adaptability to complex information environment and emergencies, realize intelligent information processing, simplify the information processing procedure, and improve the processing efficiency and quality.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and in particular to an information data processing method, apparatus, and electronic device. Background Technology

[0002] In related technologies, the data volume in the information environment is massive and growing rapidly. The hardware performance and software algorithms of existing systems may not be able to meet the requirements for rapid processing and analysis of large amounts of data. For example, when faced with sudden information outbreaks, the data volume can surge instantly. If the system's storage and computing capabilities are insufficient, the processing speed will be slow. Furthermore, in the current information dissemination process, uniform content and methods are often used, making it impossible to understand the feedback and changing needs of different groups in a timely manner. Dissemination strategies are difficult to optimize and adjust based on actual results, and cannot be adjusted to the differences in information acceptance and preferences among different groups. For example, young people may prefer intuitive information formats such as short videos and pictures, while older people may prefer detailed text descriptions. However, existing systems do not adjust to these differences, resulting in poor information dissemination effectiveness.

[0003] Therefore, there is an urgent need to propose a completely new technical solution to solve at least one technical problem in the relevant technologies. Summary of the Invention

[0004] This application addresses the technical problems existing in the prior art by providing an information data processing method, apparatus, and electronic device to solve the technical problems of slow processing speed and poor information dissemination effect in the prior art.

[0005] In a first aspect, embodiments of this application provide an information data processing method, the method comprising:

[0006] Acquire information data to be analyzed; the information data comes from different social media platforms;

[0007] A hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of hot topics in the dynamic graph of social networks to obtain information analysis results of hot topics; the information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies;

[0008] Based on the information analysis results and the types of various social media platforms, real-time interaction strategies corresponding to trending topics are generated; the real-time interaction strategies include at least: opinion leader interaction strategies adapted to activity information and social relationship influence, and topic information release strategies adapted to the characteristics of information changes;

[0009] The real-time interaction strategy is executed through a real-time interaction model to simulate the behavioral characteristics of users on various social media platforms to obtain matching information to be published and / or behavioral operation instructions, and push them to users for confirmation of whether to execute, thereby achieving positive information guidance on trending topics.

[0010] Secondly, embodiments of this application provide an information data processing apparatus, which includes at least the following units:

[0011] The acquisition unit is configured to acquire information data to be analyzed; the information data comes from different social media platforms.

[0012] The identification unit is configured to use a hot topic detection model to identify hot topics in the information data in real time, and track the changing trends of hot topics in the dynamic graph of social networks to obtain information analysis results of hot topics; the information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies;

[0013] The generation unit is configured to generate real-time interaction strategies corresponding to trending topics based on the information analysis results and the types of various social media platforms. The real-time interaction strategies include at least: opinion leader interaction strategies adapted to activity information and social relationship influence, and topic information release strategies adapted to the characteristics of information changes.

[0014] The interaction unit is configured to execute the real-time interaction strategy through a real-time interaction model on various social media platforms to guide positive information about trending topics. The real-time interaction model is used to simulate the behavioral characteristics of users on various social media platforms to obtain information to be published or operation instructions, and push them to users for confirmation of whether to execute.

[0015] Thirdly, embodiments of this application provide an electronic device, the electronic device comprising:

[0016] At least one processor and a memory; wherein the memory is used to store a computer program, and the processor is used to invoke the computer program stored in the memory to execute the information data processing method of the first aspect.

[0017] Fourthly, a computer-readable storage medium is provided, comprising instructions that, when executed on a computer, cause the computer to perform the information data processing method of the first aspect.

[0018] The beneficial effects of this application are: it provides an information data processing method, apparatus, and electronic device. In this technical solution, firstly, information data to be analyzed is acquired. Then, a hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of these hot topics in the dynamic graph of social networks, obtaining information analysis results for the hot topics. The information analysis results include at least: the information change characteristics of the hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies. Next, based on the information analysis results and the type of each social media platform, a real-time interaction strategy corresponding to the hot topic is generated. The real-time interaction strategy includes at least: an opinion leader interaction strategy adapted to activity information and social relationship influence, and a topic information publishing strategy adapted to the information change characteristics. Finally, the real-time interaction strategy is executed through the real-time interaction model to simulate the behavioral characteristics of users on various social media platforms to obtain matching information to be published and / or behavioral operation instructions, and pushes these to the user for confirmation, thereby achieving positive information guidance for the hot topics. In summary, this solution rapidly senses and responds to information changes through distributed data acquisition, real-time analysis models, and automated strategy generation. Leveraging in-depth analysis through multi-technology integration and targeted strategy formulation, it accurately grasps information, enhances guidance effectiveness, and flexibly adjusts strategies based on different platform characteristics and dynamic information. This strengthens its adaptability to complex information environments and unforeseen events, achieving full automation and intelligence in the information processing process, simplifying workflows, and improving processing efficiency and quality. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating an information data processing method according to an embodiment of this application;

[0020] Figure 2 This is a schematic diagram of the structure of an information data processing device according to an embodiment of this application;

[0021] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application;

[0022] Figure 4 This is a schematic diagram of the structure of a medium according to an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] To address at least one technical problem in the related technologies, embodiments of this application provide an information data processing method, apparatus, and electronic device.

[0025] In this application's technical solution, the information data to be analyzed is first acquired; this information data comes from different social media platforms. For example, a distributed crawler system is used to crawl data from multiple platforms, and the data is stored in a distributed database after data cleaning and structuring. This distributed architecture improves data collection efficiency, enabling the rapid acquisition of massive amounts of information data to meet real-time analysis needs; data cleaning and structuring provide high-quality, standardized data for subsequent analysis, reducing noise interference, minimizing analysis errors, and ensuring that subsequent steps are based on reliable data. Furthermore, a hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of these hot topics in the social network dynamic graph, obtaining the information analysis results of the hot topics. The information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships. The hot topic detection model is obtained by combining graph neural networks and deep learning. The information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies. Thus, by using a hot topic detection model combining graph neural networks and deep learning, hot topics are identified in real time, their changes in the social network dynamic graph are tracked, and the information change characteristics, opinion leader activities, and social relationship influence are analyzed. Here, graph neural networks excel at processing social network structure data, accurately capturing the relationships between nodes (users, topics, content, etc.). Combined with the powerful feature extraction and pattern recognition capabilities of deep learning, they can quickly and accurately identify trending topics. By tracking the dynamic graph of social networks, the spread path, speed, and scope of trending topics can be comprehensively grasped. Analysis of the influence of opinion leaders and social relationships helps to identify key nodes and forces in information guidance, providing a strong basis for subsequent strategy formulation. Next, based on the information analysis results and the type of each social media platform, real-time interaction strategies corresponding to trending topics are generated. These real-time interaction strategies include at least: opinion leader interaction strategies adapted to activity information and social relationship influence, and topic information release strategies adapted to the characteristics of information changes. Here, based on the information analysis results and the type of social media platform, real-time interaction strategies including opinion leader interaction strategies and topic information release strategies are generated. Thus, strategies are formulated according to the characteristics of different platforms (such as social dissemination attributes, the authority of news websites) and the characteristics of information changes (such as dissemination trends, public sentiment), making information dissemination more aligned with the habits and needs of platform users and improving dissemination effectiveness. Developing interaction strategies tailored to the activities and influence of opinion leaders can effectively leverage their power to guide information, enhancing the relevance and effectiveness of information guidance and avoiding the waste of resources and poor results caused by a one-size-fits-all approach. Finally, the real-time interaction strategy is executed through a real-time interaction model to simulate user behavior characteristics across various social media platforms, obtaining matching information to be published and / or behavioral instructions, which are then pushed to users for confirmation, thereby achieving positive information guidance on trending topics.Here, a real-time interactive model simulates user behavior characteristics to obtain information to be published or operational instructions, which are then pushed to the user for confirmation and execution. Thus, the real-time interactive model simulates real user behavior, generating information and instructions that better align with the platform's user behavior patterns, making information guidance more natural and readily accepted. The push confirmation mechanism empowers users with final decision-making authority, ensuring the direction of guidance while increasing the flexibility and reliability of strategy execution and reducing the risk of misguidance.

[0026] In summary, the embodiments of this application rapidly perceive and respond to information changes through distributed data acquisition, real-time analysis models, and automated strategy generation. These embodiments can flexibly adjust strategies based on different platform characteristics and dynamic information, enhancing adaptability to complex information environments and unexpected events, achieving full automation and intelligence in information processing, simplifying information processing procedures, and improving processing efficiency and quality.

[0027] The technical solution and information data processing scheme provided in this application can also be executed by an electronic device, which can be a server, server cluster, or cloud server. The electronic device can also be a terminal device such as a mobile phone, computer, tablet computer, wearable device, or dedicated device (such as a dedicated terminal device with an information data processing method system). These electronic devices can also incorporate the chips or other hardware processing units described in the above embodiments. Alternatively, these electronic devices can also install a service program for executing the information data processing scheme.

[0028] Figure 1 A flowchart illustrating an information data processing method provided in this application embodiment is shown below. Figure 1 As shown, the method includes the following steps:

[0029] 101. Obtain the information data to be analyzed;

[0030] 102. A hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of hot topics in the dynamic graph of social networks to obtain the information analysis results of hot topics;

[0031] 103. Based on the information analysis results and the type of each social media platform, generate real-time interaction strategies corresponding to trending topics;

[0032] 104. The real-time interaction strategy is executed through a real-time interaction model to simulate the behavioral characteristics of users on various social media platforms to obtain matching information to be published and / or behavioral operation instructions, and push them to users for confirmation of whether to execute, thereby achieving positive information guidance on hot topics.

[0033] In this embodiment, the information data comes from different social media platforms.

[0034] For example, social media platforms offer rapid information dissemination. Users can quickly share information through short texts, images, and videos, and leverage forwarding functions to achieve widespread information sharing. They possess powerful topic and trending search functions, reflecting social hotspots and information trends in real time, making them important platforms for the public to access news and participate in discussions. Short video platforms, with short videos as their primary content format, offer broad entertainment, encompassing music, dance, humor, and other lighthearted content. These platforms provide a rich variety of creation tools and effects, making it easy for users to create creative videos, and the short video length aligns with the fast-paced lifestyle of modern people.

[0035] In this embodiment, the information analysis results include at least: the information change characteristics of trending topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships. Further optionally, the trending topic detection model is obtained by combining graph neural networks and deep learning.

[0036] The characteristics of information changes of trending topics on various social media platforms are as follows: different social media platforms have different user groups, dissemination mechanisms and content characteristics, and the characteristics of information changes of trending topics on each platform are also different.

[0037] Social media platforms offer rapid and wide-reaching dissemination. Their openness and immediacy allow topics to quickly gain momentum, spreading like a virus. Users can share information quickly via short texts, images, and videos, and the forwarding function enables information to circulate rapidly among a large user base, generating widespread attention and discussion. In contrast, articles published on news platforms spread in a layered, circle-like fashion. Based on user relationships, articles first circulate within the followers' circles, triggering interaction and sharing from some users, and then penetrating other circles within their social networks. This dissemination method results in a relatively stable rise in popularity; however, with in-depth content, it can create a lasting and profound impact on the target audience.

[0038] Short video platforms rely on algorithmic recommendation mechanisms, which can lead to explosive growth in the popularity of topics. These platforms accurately analyze user interests and preferences based on data such as browsing history, likes, and comments, pushing relevant content to potential users. Once a topic resonates with a large number of users, it can gain massive exposure in a short period, with a surge in views, likes, comments, and other metrics, resulting in a rapid increase in popularity.

[0039] On social media platforms, public emotions are expressed more directly and in greater variety. Due to the relative freedom of speech, users often express their inner thoughts without thinking, which may result in a large number of emotional comments. Both positive and negative emotions can quickly converge, forming a powerful wave of information. Emotions spread rapidly and are easily contagious and amplified.

[0040] Users on knowledge-based platforms tend to be more rational in their emotions. Their user base is mostly there to seek knowledge and discuss issues. When faced with a topic, they tend to approach it from the perspectives of professional knowledge and logical analysis, participating in discussions by presenting facts and reasoning. Even when different viewpoints clash, they mostly remain within a rational framework, with fewer instances of extreme emotional expression.

[0041] On lifestyle sharing platforms, public sentiment tends to be positive. These platforms focus on everyday life, product recommendations, and personal experiences, where users share beautiful moments and product reviews, creating a warm and positive atmosphere. The content posted and interacted with largely revolves around pleasant things, conveying a love for life and a positive attitude.

[0042] Users on different platforms may have different emotional inclinations towards the same trending topic. For example, when it comes to consumer-related topics, the emotional inclination in the comments section of e-commerce platforms is related to product reviews, while on social media platforms, users may pay more attention to brand image and social impact.

[0043] Information on opinion leaders' activities includes the quantity, frequency, and interaction (such as likes, comments, and shares) of content they post related to trending topics. For example, if a tech-focused opinion leader frequently posts reviews of a new phone on social media and receives a high level of interaction, it indicates a high level of engagement with the trending topic.

[0044] The social influence of opinion leaders can be analyzed using graph neural networks to examine their connections and locations within social networks. Highly influential opinion leaders may connect with multiple different social groups, enabling their opinions to spread and influence others on a wider scale.

[0045] Trending topic detection models treat users, topics, and content on social media platforms as nodes, and user interactions and relationships between topics and content as edges, constructing a dynamic graph of the social network. Graph neural networks can learn the feature representations of nodes and edges, uncovering latent structural information within the social network. For example, by analyzing the degree centrality and betweenness centrality of nodes, important user and topic nodes can be identified. As the dynamic graph of the social network changes over time, graph neural networks can track these changes in real time and discover the development trends of trending topics. Deep learning models utilize convolutional neural networks (CNNs), recurrent neural networks (RNNs), and their variants (such as LSTM and GRU) to extract features from text, images, and video data. For example, CNNs can extract local features from text, while RNNs are suitable for processing sequential data and can capture the contextual information of the text. Deep learning models can be used to classify trending topics, determining their type and sentiment. Furthermore, they can predict the development trends of trending topics, such as increases or decreases in popularity and potential related topics.

[0046] In this way, combining graph neural networks and deep learning can fully leverage the advantages of both. Graph neural networks process the structural information of social networks, while deep learning processes the content information of data, thereby more comprehensively and accurately identifying trending topics, tracking their changing trends, and improving the accuracy and reliability of information analysis.

[0047] Alternatively, the characteristics of information change may include at least: information dissemination trends, public sentiment, and emotional tendencies.

[0048] Information dissemination trends refer to the dynamic changes and patterns exhibited by information during its dissemination. This includes the speed, scope, and path of information dissemination, as well as the phased characteristics of its spread. For example, in certain trending events, information may spread rapidly through platforms such as social media within a short period, exhibiting an exponential growth trend, and its reach may quickly expand from specific groups to the entire society. As time progresses, the information's popularity may gradually decline, showing a decaying trend. By analyzing information dissemination trends, we can understand how information spreads among different groups and platforms, thereby grasping the rhythm and direction of information development and providing a basis for information response and guidance.

[0049] Public sentiment refers to the overall emotional state and emotional response of the public to a specific event, phenomenon, or topic. Public sentiment is diverse and complex, with common emotions including anger, joy, sadness, anxiety, worry, dissatisfaction, and support. The emergence of public sentiment is often influenced by the nature and impact of the event itself, as well as the public's own interests and values. For example, in the face of natural disasters, the public may express concern and sympathy; while regarding some unjust social phenomena, the public may express anger and dissatisfaction. Public sentiment directly reflects the public's attitudes and feelings towards events and plays a crucial role in driving the development and evolution of information.

[0050] Emotional inclination primarily refers to the emotional tone and value orientation inherent in how the public expresses their views and attitudes towards a particular issue. It is deeper and more concrete than mere public emotion, encompassing not only simple emotional expression but also the public's evaluation and judgment of the matter. Emotional inclination can be categorized into three types: positive, negative, and neutral. Positive emotional inclination manifests as approval, support, and appreciation of the issue; negative emotional inclination is reflected in denial, opposition, and criticism; and neutral emotional inclination indicates that the public holds an objective and rational attitude towards the issue, without a significant emotional bias. For example, regarding a newly launched product, consumers may express a liking or disliking emotional inclination based on their own user experience and feelings. This emotional inclination influences other consumers' views and purchasing decisions regarding the product, and also significantly impacts product-related information.

[0051] In this embodiment of the application, the real-time interaction strategy includes at least: an opinion leader interaction strategy adapted to activity information and social relationship influence, and a topic information release strategy adapted to the characteristics of information change.

[0052] The opinion leader interaction strategy, centered on event attributes and opinion leader influence, achieves efficient allocation of information guidance resources through precise screening, tiered cooperation, and dynamic maintenance. In the screening phase, KOLs are matched based on event type and objectives. For example, technology review KOLs are invited to interpret product technical highlights for new product launches, while mid-tier KOLs in the beauty vertical are selected for targeted marketing to promote beauty products. For brand awareness enhancement, top-tier KOLs are used to amplify the reach. A tiered strategy is employed in cooperation: high-influence KOLs are deeply engaged, invited to participate in product development and co-create exclusive content; mid-influence KOLs are used to stimulate participation from their respective user groups through topic challenges and interactive discussions; and low-influence KOLs are incentivized with coupons and traffic support to build a broad communication matrix. Furthermore, by monitoring the effectiveness of KOL posts in real time, timely intervention is provided for negative information, and KOL profiles are established for long-term relationship maintenance, ensuring the sustainability of cooperation and the stability of information guidance.

[0053] The information dissemination strategy dynamically adjusts the content and form of information releases based on information dissemination trends, public sentiment, and emotional tendencies. In terms of dissemination trends, during the information explosion phase, authoritative interpretations are released frequently to seize the information high ground; during the diffusion phase, multi-platform features are utilized for coordinated dissemination; and during the decline phase, results are reviewed to consolidate influence. Regarding public sentiment, positive emotions are reinforced by showcasing successful cases to evoke emotional identification; negative emotions are addressed with reassuring statements and solutions to resolve conflicts; and neutral emotions are addressed through interactive formats such as polls and Q&A sessions to activate participation. In terms of emotional tendencies, positive evaluations are amplified through activities such as "likes, shares, and prize draws." A FAQ page is created to address negative viewpoints; comprehensive and objective product reviews are provided to neutral groups to guide them in making rational judgments, ultimately achieving precise and effective information guidance.

[0054] In step 101, the system acquires the information data to be analyzed. Specifically, it can scrape data from multiple social media platforms, news websites, and various forums in parallel. Simultaneously, it employs appropriate data parsing techniques for different platform data formats (such as JSON and HTML), such as using regular expressions, XPath, or CSS selectors to extract the required information. For data storage, the scraped data can be stored in a distributed database, such as the Hadoop Distributed File System (HDFS) or a distributed key-value store (such as Redis), for subsequent large-scale data processing and analysis.

[0055] On the other hand, many social media platforms and news websites provide official APIs (Application Programming Interfaces). By calling these APIs, data can be retrieved according to the platform's specifications. When using APIs to retrieve data, it's necessary to understand the platform's API documentation, including the interface parameter requirements and the returned data format. To improve the efficiency of data retrieval, asynchronous programming techniques can be used to initiate multiple API requests simultaneously, reducing waiting time.

[0056] Optionally, raw data obtained from the internet often contains a large amount of noise and useless information, requiring cleaning and preprocessing. Data cleaning includes removing duplicate data, handling missing values ​​and outliers, etc. For text data, preprocessing such as word segmentation, part-of-speech tagging, and named entity recognition is also required for subsequent semantic analysis. When processing image and video data, image recognition and video content analysis may be necessary to extract key information. To improve the efficiency of data cleaning and preprocessing, some open-source data processing tools and frameworks can be used, such as Apache Spark and Apache Flink. These tools can handle large-scale data and provide rich data processing functions and algorithms.

[0057] To obtain more comprehensive information, it is necessary to integrate data from different data sources. This requires standardized format conversion and processing of different data types. For example, text, image, and video data need to be converted into a unified feature representation for subsequent analysis and mining. During the integration process, it is also necessary to address data consistency and conflict issues to ensure the accuracy and reliability of the integrated data. To achieve multi-source data fusion, various data fusion algorithms and models can be used, such as feature-based fusion algorithms and decision-based fusion algorithms.

[0058] Information data is time-sensitive; to understand information dynamics promptly, real-time data acquisition and monitoring are necessary. Real-time stream processing technologies, such as Apache Flink and Apache Kafka, can be used to process and analyze real-time data. By setting real-time monitoring metrics, such as information popularity and sentiment, alerts can be issued promptly when these metrics exceed certain thresholds. Simultaneously, to ensure the accuracy and integrity of real-time data, real-time verification and correction are required. During real-time data acquisition and monitoring, system performance and scalability must also be considered to ensure the system can handle large-scale real-time data.

[0059] Specifically, in 101, a distributed crawler system is built, which consists of multiple crawler nodes that can be distributed across different servers to crawl data from platforms such as social media and news websites in parallel.

[0060] Specific data cleaning and preprocessing strategies were developed for different platforms. For social media data, due to the casual nature of user expression, there may be numerous typos, emojis, and special characters, requiring text standardization processing, such as removing special characters and converting emojis into text descriptions. Additionally, social media data may contain duplicate content, necessitating the removal of duplicate data using deduplication algorithms. For news website data, there may be inconsistent formatting issues, such as date formats and author information formats, requiring format standardization. Furthermore, both social media and news website data may contain missing values. For handling missing values, methods such as deletion or imputation (e.g., using the mean, median, or mode) can be chosen depending on the specific situation. For example, if important information is missing from a data record and cannot be supplemented through other means, the record can be deleted; while for missing minor information, reasonable default values ​​can be used for imputation.

[0061] Then, the cleaned and preprocessed data is stored in a distributed database. Distributed databases offer advantages such as high availability, scalability, and high performance, meeting the storage and management needs of large amounts of information. When choosing a distributed database, the appropriate database type can be selected based on the characteristics of the data and the application scenario, such as key-value stores (e.g., Redis), document databases (e.g., MongoDB), or relational databases (e.g., distributed versions of MySQL). When storing data, a reasonable data table structure or data model is designed according to the data type and purpose. For example, for social media data, corresponding data tables can be created according to different entities such as users, posts, and comments, and appropriate indexes can be established to improve data query efficiency. For news website data, it can be stored according to dimensions such as news category and publication time, facilitating subsequent analysis and retrieval.

[0062] By operating the above data acquisition modules, we can obtain rich information data from multiple platforms, and effectively clean, preprocess, and store this data to provide a high-quality data foundation for subsequent information analysis and processing.

[0063] As an optional embodiment, after obtaining the information data to be analyzed in step 101, the user's task requirements for the positive information guidance process can also be obtained; based on the task requirements and the information data, a guidance decision model adapted to the hot topics in the information data is constructed using a big oracle model; the guidance decision model includes a behavior suggestion module and a long-term monitoring module; through the behavior suggestion module, the changing trends of the task requirements and the information data are predicted to obtain monitoring behavior suggestions for hot topics; through the long-term monitoring module, the monitoring methods of the task requirements, the information data, and the monitoring behavior suggestions are matched to obtain a long-term monitoring strategy for hot topics; the long-term monitoring strategy is executed on various social media platforms, and the changes in information data after the execution of the real-time interaction strategy on each social media platform are monitored, and the information release strategy and behavior operation strategy in the real-time interaction strategy are dynamically adjusted.

[0064] Specifically, after obtaining the information data to be analyzed, the task requirements for guiding the positive information process are further obtained from the user.

[0065] Furthermore, based on the acquired task requirements and information data, a guidance decision-making model adapted to the hot topics in the information data is constructed using a large language model. Through learning from a large amount of text data, the large language model can understand the semantic information in the task requirements and information data, and build the model based on this information. This model includes a behavior suggestion module and a long-term monitoring module.

[0066] Next, the behavior suggestion module conducts in-depth analysis of the task requirements and information data. Utilizing a large language model to learn from historical data of similar information events and understand current data, it predicts changing trends. For example, the analysis reveals that if effective measures are not taken over time, negative information may spread further, affecting more related fields and user groups. Based on this, the behavior suggestion module generates monitoring behavior suggestions for trending topics. These might include inviting authoritative industry experts to publish objective analysis articles on product quality issues, emphasizing product improvement measures and quality enhancement directions; and arranging for official brand customer service personnel to promptly and professionally respond to negative user comments on social media platforms, explaining the causes of the problems and solutions.

[0067] Then, the long-term monitoring module comprehensively analyzes the task requirements, information data, and monitoring behavior suggestions to match appropriate monitoring methods and obtain a long-term monitoring strategy for trending topics. For different social media platforms, different monitoring frequencies and indicators are formulated based on their data characteristics and user behavior patterns. On platforms primarily featuring text-based discussions, the module monitors the popularity, number of comments, sentiment, and dissemination scope of topics related to the brand's products hourly. On short video platforms, it monitors the real-time views, likes, comments, and user interaction behavior of relevant videos. Simultaneously, the module monitors the effectiveness of measures proposed by the behavior suggestion module, such as the readership and reposting of expert articles and their impact on the sentiment of the information.

[0068] Finally, based on the long-term monitoring strategy, corresponding monitoring operations were performed on various social media platforms. During the monitoring process, close attention was paid to changes in information data after the implementation of the real-time interaction strategy on each social media platform. If it was found that negative information was not effectively controlled after the implementation of the real-time interaction strategy, or even showed a further deterioration trend, such as a continuous increase in the number of negative comments on platforms mainly based on text-based discussions, and an increasingly strong negative sentiment, then the information release strategy and behavioral operation strategy within the real-time interaction strategy were dynamically adjusted based on the monitored information data changes. For example, the information release strategy was adjusted to increase the frequency and coverage of positive information releases, and to publish more actual cases of product quality improvement and positive user feedback; the behavioral operation strategy was adjusted to strengthen the handling of negative comments, not only by replying and explaining, but also by providing some compensation measures or promotional activities to appease users' emotions and guide information towards a positive direction.

[0069] Through the above steps, from obtaining task requirements to building a guided decision-making model, to the operation of each module and the final strategy adjustment, a complete and intelligent positive information guidance process is formed to cope with the complex and ever-changing information environment and achieve effective guidance and management of information.

[0070] Further, optionally, in the above steps, the long-term monitoring strategy is implemented on each social media platform, including:

[0071] Based on the aforementioned long-term monitoring strategy, a long-term information evaluation system matching trending topics is constructed. Using this system, information change predictions are made based on long-term information monitoring data collected from various social media platforms and the information data changes after implementing the real-time interaction strategy, yielding information change index data. A long-term social network dynamic graph is constructed from the long-term information monitoring data, the information change index data, and the information data changes. The degree centrality and betweenness centrality of nodes in the dynamic graph are calculated to assess the social relationship influence of opinion leaders, thus obtaining dynamic information change data corresponding to the long-term social network. Based on the information change index data and the dynamic information change data, a long-term information evaluation report for trending topics is generated.

[0072] Specifically, firstly, based on a long-term monitoring strategy and considering the characteristics and needs of trending topics, a long-term information evaluation system is constructed using technologies such as natural language processing and machine learning. For example, for the trending topic of "the application of artificial intelligence in the medical field," the system will focus on relevant professional terminology, industry trends, and public discussion. Through learning and analyzing a large amount of historical information data, the dimensions and indicators for evaluation are determined, such as the topic's popularity, sentiment, dissemination scope, and trends in public attention, thereby constructing a system capable of accurately evaluating the information status of this trending topic.

[0073] Furthermore, using the established long-term information evaluation system, the system analyzes the long-term information monitoring data collected from various social media platforms and the changes in information data after implementing real-time interaction strategies. Utilizing time series analysis and predictive algorithms in machine learning, such as the ARIMA model and neural networks, the system models and predicts the changing trends of information data, obtaining information change indicator data. These indicators may include the magnitude of increase or decrease in topic popularity over a future period, the proportion of changes in positive or negative sentiment, and the expansion or contraction of the dissemination scope.

[0074] Next, by integrating long-term information monitoring data, information change indicator data, and information data changes, a dynamic graph of the long-term social network is constructed, using users, topics, and events on social media platforms as nodes and user interactions (such as following, commenting, and forwarding) and topic relationships as edges. For example, on platforms primarily featuring text-based discussions, using different user accounts as nodes, if user A frequently comments on and forwards user B's content related to "the application of artificial intelligence in the medical field," then an edge is established between A and B, indicating that they have an interactive relationship on this topic.

[0075] The social influence of opinion leaders is assessed by calculating the degree centrality and betweenness centrality of nodes in a dynamic graph. Degree centrality refers to the number of direct connections between a node and other nodes; nodes with high degree centrality typically have broader social connections and can spread information quickly. Betweenness centrality measures how much a node acts as a mediator between other nodes in the network, acting as the shortest path. Nodes with high betweenness centrality play a crucial bridging role in information dissemination and have greater control over the flow and spread of information. These two indicators can be used to identify opinion leaders with significant influence in the dissemination of trending topics.

[0076] Based on the assessment results of the influence of opinion leaders' social relationships, combined with information monitoring data and change indicator data, dynamic information change data corresponding to long-term social networks were obtained. This data reflects the trends in the spread and evolution of information on social networks under the influence of opinion leaders. For example, a positive comment posted by an opinion leader may influence their audience's attitude towards the topic to shift in a positive direction, thus causing corresponding changes in information data.

[0077] Finally, based on information change indicator data and dynamic information change data, a long-term information evaluation report on hot topics is generated according to a certain format and structure. The report content may include an overall overview of the information, trend analysis of information changes, assessment of the influence of opinion leaders, predictions of future information development, and corresponding suggestions and countermeasures. For example, a long-term information evaluation report on the hot topic of "the application of artificial intelligence in the medical field" would summarize the dissemination of this topic on social media platforms over a period of time, analyze whether the information is developing in a positive or negative direction, identify key opinion leaders and their guiding role in information dissemination, predict potential future information hotspots and trends, and provide relevant institutions or enterprises with suggestions on how to better guide information and improve public awareness and acceptance of the technology. The report can employ data visualization methods, such as using charts and graphs to display information data and analysis results, so that decision-makers can intuitively understand the information status and development trend.

[0078] Optionally, in the above steps, monitoring the changes in information data after the execution of the real-time interaction strategy on various social media platforms, and dynamically adjusting the information publishing strategy and behavioral operation strategy in the real-time interaction strategy, including:

[0079] Extract information change data after the execution of the real-time interaction strategy from various social media platforms; determine the speed and scope of information dissemination trends on each social media platform based on the information change data; and dynamically adjust the frequency and scope of information release for hot topics according to the speed and scope of information dissemination trends.

[0080] For example, on various social media platforms, data collection technologies (such as web crawlers and API calls) are used to extract data related to the content following the execution of real-time interaction strategies. Specifically, data such as posts, comments, likes, and reposts related to trending topics, as well as relevant user behavior information, are collected. For instance, on social media platforms, all blog posts with specific hashtags published after the execution of real-time interaction strategies are extracted, including the content, publication time, publisher information, number of likes, number of comments, and number of reposts; on short video platforms, data such as the number of views, likes, comments, and user interaction behavior of relevant short videos are obtained. This data comprehensively reflects the changes in information about trending topics on various platforms after the execution of real-time interaction strategies.

[0081] Furthermore, based on the extracted information change data, data analysis and mining techniques are used to determine the speed and scope of information dissemination trends. The speed of change is determined by analyzing the increase or decrease of data related to trending topics over a period of time (e.g., hourly, daily). For example, comparing the change in the volume of discussion (the sum of blog posts, comments, and reposts) of related topics before and after implementing the real-time interaction strategy, if the discussion volume increases rapidly after implementing the strategy, it indicates that the information dissemination speed has accelerated; conversely, the dissemination speed has slowed down. The scope of dissemination is considered from multiple dimensions, including the range of user groups involved (e.g., changes in the number of users of different ages, regions, and professions), the cross-platform dissemination of the topic (whether it spreads from one platform to other platforms), and the relevance of related topics (whether it has sparked more discussions on related topics). For example, if a trending topic that was originally discussed only among users in the technology field gradually spread to the general public after implementing the strategy and sparked related discussions on multiple social media platforms, it indicates that the scope of dissemination has expanded.

[0082] Next, based on the speed and scope of changes in the determined information dissemination trend, the frequency and scope of information releases for trending topics are dynamically adjusted. When the speed of information dissemination accelerates and the scope expands, the frequency of information releases is appropriately increased to maintain the topic's popularity and attention. For example, on platforms primarily focused on text-based discussions, where two official announcements about trending topics are released daily, this can be increased to four or more per day, with more attractive content and formats chosen (such as videos or images paired with detailed text descriptions). Simultaneously, the scope of information releases is expanded beyond existing channels and platforms to other relevant platforms or channels, attracting more user attention. For instance, in addition to releasing information, related short videos or live streams can be created and promoted on short video platforms or lifestyle sharing platforms. Conversely, when the speed of information dissemination slows and the scope shrinks, the frequency of information releases is reduced to avoid excessive dissemination that could alienate users. At the same time, the content of the released information is optimized to be more targeted and attractive, thus reigniting user interest. For example, information is screened and integrated, highlighting key points and highlights, and released during periods of high user activity. By making such dynamic adjustments, the information dissemination strategy in the real-time interaction strategy can better adapt to the actual changes in information, thereby improving the effectiveness of information guidance.

[0083] Optionally, in step 102, a hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of hot topics in the social network dynamic graph to obtain information analysis results of hot topics, including:

[0084] A trending topic detection model is used to perform semantic analysis on the information data to extract event keywords, event themes, and sentiment trends. Based on the extracted event keywords, event themes, and sentiment trends, user nodes, topic nodes, and content nodes are constructed, and these nodes are connected to form a dynamic social network graph. Event identification is performed based on the dynamic social network graph to obtain the trending topics to be monitored. Trending topics are determined by the access popularity of event nodes, the number of associated edges, and the relevant information density in the dynamic social network graph. Opinion leader activity analysis is performed based on the user nodes in the dynamic social network graph to obtain key opinion leaders related to the trending topics and their influence on the topics. The system analyzes relevant activity information and social relationship influence; it uses graph neural networks (GNNs) to analyze the degree centrality, betweenness centrality, and PageRank value of each user node to identify key opinion leaders; it extracts user nodes, topic nodes, and content nodes related to hot topics from the social network dynamic graph to construct a local social network dynamic graph corresponding to the hot topic; it selects a topic analysis model matching the hot topic from a candidate data model library; and it uses the topic analysis model to perform adaptive topic analysis on the local social network dynamic graph to obtain the information analysis results; wherein, the topic analysis model is pre-trained using historical information data of historical hot topics matching the hot topic.

[0085] In this embodiment, a user node can be considered as each user on the social platform. These users can be ordinary users, key opinion leaders (KOLs), event organizers, or other related organizations. Each user node has a unique identifier, such as a user ID or username, and may also contain other attribute information about the user, such as the number of other users following the user, the following list, and the user's activity level, in order to analyze the user's influence and behavioral patterns later.

[0086] Topic nodes can be defined as each trending topic related to an activity identified from the information data. Topic nodes can be identified using key information about the topic, such as topic tags and subject content. Adding attributes to topic nodes, such as the topic's popularity score (calculated based on the number of related posts, interaction volume, etc.) and the topic's sentiment tendency (derived through sentiment analysis of related texts), helps in understanding the topic's characteristics and development trend.

[0087] Content nodes can be used to identify each post, article, comment, or other content published on a social media platform. A content node's identifier can be a unique ID, and its attributes can include the content's publication time, length, type (e.g., text, image, video), and keywords, allowing for analysis of the impact of different content on the spread of trending topics.

[0088] Based on the three types of nodes mentioned above, edges between users can be established as follows: if one user follows another, a directed edge is created between these two user nodes, pointing from the follower to the followed user. This type of edge represents the following relationship between users, reflecting the potential path of information dissemination among them. Furthermore, if two users have interacted, such as commenting, liking, or sharing, an edge can also be established between them. The weight of the edge can be set according to factors such as the frequency and intensity of the interaction. For example, sharing may be given a higher weight than liking, to reflect the different impacts of different interaction methods on information dissemination.

[0089] The edge between a user and a topic can be created when a user posts content related to a trending topic. A directed edge is established between the user node and the corresponding topic node, with the direction pointing from the user to the topic. The weight of the edge can be determined based on the contribution of the user's content to the topic's popularity. For example, if the posted content generates a large amount of discussion and sharing, then the edge has a relatively high weight, indicating that the user played a significant role in spreading the topic.

[0090] The edge between users and content can be defined as follows: When a user publishes content, a directed edge is established between the user node and the content node, pointing from the user to the content, representing the content creator. When a user comments on, likes, or shares content, a corresponding directed edge is also established, with the direction determined by the specific action. For example, the comment edge points from the user to the content, while the share edge points from the content to the user who shared it, thus describing the interactive relationship between users and content.

[0091] The edges between topics and content can be structured as follows: if a piece of content relates to a trending topic, a directed edge is created between the topic node and the content node, pointing from the topic to the content, indicating that the content is about that topic. The weight of the edge can be set according to the relevance between the content and the topic. For example, edges that directly mention topic keywords and discuss them in detail have higher weights, while edges that only briefly mention the topic have lower weights. By constructing these edges, the distribution and dissemination of related content under each trending topic can be clearly observed.

[0092] Based on the foregoing, in section 102, firstly, the large language model within the trending topic detection model is used to perform semantic analysis on the information data. Word segmentation technology is used to break the text data down into individual words, part-of-speech tagging determines the part of speech of each word, and dependency parsing clarifies the grammatical relationships between words, thereby extracting event keywords, event themes, and sentiment tendencies. Based on this extracted information, user nodes, topic nodes, and content nodes are constructed. Then, according to the relationships between nodes, such as a user posting content related to a certain topic, the nodes are connected to form a dynamic social network graph, visually demonstrating the connections between different elements in the social network.

[0093] Event identification is performed based on a constructed social network dynamic graph. By analyzing the characteristics of nodes and edges in the graph, such as node attributes (keywords, topics, etc.), edge connections, and weights, the hot topics to be monitored are determined. Specifically, hot topics are determined by comprehensively considering the access popularity of event nodes (e.g., views and clicks of content related to a topic), the number of associated edges (the number of connections generated by user interactions and content dissemination related to the topic), and the relevant information density in the social network dynamic graph (the richness of information containing the topic). Topics with high access popularity, a large number of associated edges, and high information density are more likely to become hot topics.

[0094] This analysis begins with user nodes in a social network dynamic graph to identify key opinion leaders (KOLs). By collecting relevant user information, such as posted content and interaction data (likes, comments, shares, etc.), a graph neural network (GNN) is used to calculate the degree centrality (measures the number of direct connections between nodes, reflecting the node's social activity), betweenness centrality (assessing the degree to which a node acts as a mediator in the shortest path between other nodes, reflecting the node's control over information dissemination), and PageRank (an algorithm used for ranking web pages, here measuring the node's importance in the social network). This analysis identifies KOLs. After identifying KOLs, further analysis is conducted on their activities related to trending topics (such as what content they posted about trending topics and the frequency of their posts) and their social influence (the degree of impact on other users' opinions and behaviors).

[0095] This process extracts user nodes, topic nodes, and content nodes related to trending topics from the dynamic graph of a social network. These nodes are then recombined to construct a local dynamic graph of the social network corresponding to the trending topic, focusing on the network structure and information related to the trending topic. Next, topic analysis models matching the trending topic are selected from a candidate data model library. These models are pre-trained using historical information data of historical trending topics that match the trending topic, and possess analytical capabilities for specific types of trending topics. The selected topic analysis models are then used to perform adaptive topic analysis on the local dynamic graph of the social network. By combining the structural information and node characteristics in the graph, the process delves into the information change characteristics of the trending topic (such as information dissemination trends, public sentiment, and emotional tendencies), the activity information of opinion leaders, and the influence of social relationships, ultimately obtaining comprehensive and accurate information analysis results that provide strong support for subsequent information guidance and decision-making.

[0096] Optionally, in step 102, a hot topic detection model is employed to identify hot topics in the information data in real time and track the changing trends of hot topics in the dynamic graph of the social network. After obtaining the information analysis results of the hot topics, a set of hot topic indicators can be extracted from the information analysis results. In this embodiment, the set of hot topic indicators includes at least: hot topic evolution records, associated keywords, public sentiment analysis results, and opinion leader activity data; the public sentiment analysis results include at least: the proportion of each public sentiment category and its corresponding changing trend. Furthermore, the set of hot topic indicators is projected onto a visualization space and converted into dynamic data graphs, and the dynamic data graphs are bound to a visualization chart component; the dynamic data graphs include at least: a hot topic evolution graph, a public sentiment distribution graph, and an opinion leader activity distribution graph. Finally, an information data analysis report on hot topics is generated based on the set of hot topic indicators, and the information data analysis report and the visualization chart component are displayed to the user.

[0097] Specifically, after step 102, the system identifies trending topics in the information data in real time using a trending topic detection model and tracks their changing trends in the dynamic graph of social networks. After obtaining the information analysis results of the trending topics, key information is extracted from these results to form a set of trending topic indicators. Specifically, the trending topic evolution record covers the changes in topic popularity over time, clearly reflecting the entire process from the emergence, development, and gradual decline of a trending topic. Related keywords are important words closely related to the trending topic, allowing for a more accurate understanding of its core content. Public sentiment analysis results include the proportion of various public sentiment categories (such as positive, negative, and neutral) and the changing trends of these proportions over time, helping to understand the dynamic changes in public attitudes and emotional tendencies towards trending topics. Opinion leader activity data includes the number of posts, interactions (such as likes, comments, and reposts), and influence scores of opinion leaders, reflecting their role in disseminating and guiding information regarding trending topics.

[0098] Furthermore, the extracted hot topic indicators are projected onto a visualization space and converted into dynamic data charts. For the evolution records of hot topics, a hot topic evolution chart is created, typically using time as the horizontal axis and topic popularity as the vertical axis, displaying the rise and fall of topic popularity in a line chart format. Users can intuitively see the popularity of hot topics at different points in time and the trend of popularity changes. Public sentiment analysis results are converted into a public sentiment distribution chart, using bar charts or pie charts to present the proportion of each sentiment category, and heat maps to show the differences in the distribution of public sentiment in different regions, making the distribution of public sentiment clear at a glance. Opinion leader activity data is converted into an opinion leader activity distribution chart, using bar charts to compare the number of posts and interactions of different opinion leaders, combined with radar charts to show the comprehensive performance of opinion leaders in various dimensions such as influence, dissemination, and activity. Then, these dynamic data charts are bound to visualization chart components to achieve real-time linkage and updates between data and charts, ensuring that users can obtain the latest information.

[0099] Then, based on the set of trending topic indicators, an information data analysis report on the trending topics is generated. The report includes a detailed analysis of the trending topics, such as their background, development process, and current status; an interpretation of public sentiment, including the reasons for changes in the proportion of sentiment categories and the impact of public sentiment on the trending topics; and an evaluation of opinion leader activities, such as the sources of their influence and the guiding effect of their statements and behaviors on information. Simultaneously, based on the conclusions generated by the information analysis module, the development trends of the trending topics are summarized and predicted, and corresponding information guidance strategies are proposed. Finally, the generated information data analysis report and visualization chart components are presented to users. Users can intuitively understand the various indicators and changes of the trending topics through the visualization charts, and gain a deeper understanding of the connotation and impact of the trending topics through the information data analysis report, providing strong support for user decision-making. During the presentation, interactive functions are also provided, such as allowing users to preview the report content online, confirm the report or provide modification suggestions, and download the report file after confirmation. The report is also stored in the system file library for easy access and retrieval later.

[0100] Further optionally, in step 103, based on the information analysis results and the type of each social media platform, a real-time interaction strategy corresponding to the trending topics is generated, including:

[0101] A sentiment analysis model is used to identify the stance of key opinion leaders based on activity information in the information analysis results, thereby obtaining the stance type of the key opinion leaders. The stance type is one of positive support, neutral observation, or negative skepticism. A multi-dimensional influence assessment model is used to conduct multi-dimensional analysis based on the social relationship influence in the information analysis results to obtain the leader influence level of the key opinion leaders. Based on the leader influence level and the stance type, an opinion leader interaction strategy corresponding to the key opinion leader is generated. The frequency of dynamic strategy adjustment is set according to the event popularity of hot topics, and the leader influence level and the stance type are updated based on the dynamic strategy adjustment frequency. If the leader influence level and the stance type change, the opinion leader interaction strategy is switched through the rule engine bound to the social media platform. The topic content is predicted and the topic construction method is analyzed based on the characteristics of information change to obtain the topic information release strategy. The opinion leader interaction strategy and the topic information release strategy are fused to obtain the real-time interaction strategy.

[0102] Specifically, firstly, a sentiment analysis model is used to conduct an in-depth analysis of the activity information (such as content posted by key opinion leaders and discussions they participate in) in the information analysis results. Through techniques such as semantic understanding of the text and sentiment tendency judgment, the stance type of key opinion leaders on trending topics is identified. For example, if the content posted by key opinion leaders is full of affirmation, support, and positive evaluation of trending topics, then their stance type can be determined as positive support; if their comments neither clearly support nor oppose, but rather objectively state facts or raise questions, then they are determined as neutral and observing; if they express negative attitudes such as dissatisfaction, criticism, or questioning of trending topics, then they belong to the negative and questioning type.

[0103] Furthermore, a multi-dimensional influence assessment model is used to conduct a comprehensive and multi-dimensional analysis based on the social relationship influence in the information analysis results. This model considers multiple factors, such as the number of users following the key opinion leader, the activity level of these followers (interaction volume, including likes, comments, and shares), the breadth of connections within the social network (the number of connections with other important nodes), the scope and speed of information dissemination, and professional authority in specific fields. Through quantitative analysis and comprehensive evaluation of these dimensions, the influence level of the key opinion leader is determined. Influence levels may be divided into high, medium, and low levels to reflect their actual ability to disseminate trending topics and guide information on social networks.

[0104] Based on the identified levels of influence and stances of key opinion leaders, corresponding interaction strategies should be developed. For key opinion leaders who are actively supportive and have high levels of influence, a deep cooperation strategy can be adopted, such as inviting them to participate in official activities related to hot topics and jointly releasing content, to further leverage their influence to expand the reach of positive information. For key opinion leaders who are neutral and have moderate influence, detailed and accurate information can be provided to guide them to understand hot topics more deeply and try to shift their neutral stance towards a positive one. For key opinion leaders who are negative and have low influence, the focus should be on active communication and answering questions, using professional explanations and reasonable responses to alleviate their negative emotions and reduce negative impacts.

[0105] The frequency of dynamic strategy adjustments is set according to the popularity of trending topics. When a trending topic is popular and information changes rapidly, the frequency of strategy adjustments is increased, such as evaluating and adjusting it every hour or half-day; when the popularity is low and information is relatively stable, the adjustment frequency is decreased, such as once a day or once a week. Based on the dynamic strategy adjustment frequency, the leadership influence level and stance type of key opinion leaders are updated regularly. If changes in the leadership influence level and stance type are detected during the update process, such as a previously neutral key opinion leader becoming an actively supportive one, or a previously highly influential leader experiencing a decline in influence for some reason, the corresponding opinion leader interaction strategy is automatically switched through the rule engine bound to the social media platform to ensure the effectiveness and relevance of the strategy.

[0106] This analysis of topic content prediction and construction methods is based on the characteristics of information change (such as information dissemination trends, public sentiment, and emotional tendencies). By analyzing information dissemination trends, the potential future development and evolution of trending topics are predicted, thereby determining appropriate topic content. For example, if a trending topic is predicted to develop in a specific direction, relevant topic content can be prepared in advance to guide public opinion. Simultaneously, appropriate topic construction methods are selected based on public sentiment and emotional tendencies. When public sentiment is positive, positive and inspiring content is released; when public sentiment is negative, content that soothes and addresses problems is released. Based on these analytical results, a topic information release strategy is formulated, including the frequency of release, content format (such as text, images, and videos), and the selection of release platforms.

[0107] Finally, the generated opinion leader interaction strategy and topic information dissemination strategy are integrated to ensure mutual coordination and cooperation, forming an organic whole. For example, when interacting with key opinion leaders, the published topic content should align with their stance and influence to achieve optimal dissemination and information guidance effects. The resulting real-time interaction strategy can be flexibly applied across different social media platforms based on the actual situation and information changes of trending topics, enabling effective management and guidance of trending topics and improving the efficiency and quality of information response.

[0108] In one optional example, suppose a large multinational corporation hosts a global technology innovation conference, attracting numerous international participants, including technology companies, research institutions, industry experts, and technology enthusiasts from around the world. During the conference's preparation and execution, to better communicate with international participants and disseminate relevant information, the corporation employs a strategy of disseminating information using multilingual virtual avatars.

[0109] The virtual characters were designed with multiple styles and backgrounds to reflect diverse cultures.

[0110] Each virtual character is equipped with multiple language abilities, covering English, Chinese, French, Spanish, German, Japanese, and other languages. Certain language abilities are emphasized based on the character's design and target audience.

[0111] During the pre-conference promotion phase, multilingual virtual avatars released information about the conference on major international social media platforms. Simultaneously, the virtual avatars also released engaging short videos showcasing the conference preparations and venue setup, enhancing the appeal of the information.

[0112] During the conference, the virtual avatar provided real-time updates on highlights, key points from important speeches, and information from interactive sessions. For example, when a renowned international scientist delivered a keynote address, the virtual avatar quickly compiled the core arguments and key data in multiple languages ​​and published them on the relevant platforms. For interactive sessions such as online voting and Q&A, the virtual avatar guided international participants in multiple languages, ensuring that people from different language backgrounds could understand and participate. Furthermore, the virtual avatar adjusted its information delivery methods and content style to reflect the cultural characteristics of different regions.

[0113] Following the conference, the language-specific virtual avatars will release a summary and review of the event, including its achievements, future plans, and expressions of gratitude to international participants. They will also share feedback and evaluations from attendees, showcasing the conference's global impact and positive reception in multiple languages. Furthermore, the virtual avatars will guide international participants to stay informed about the companies' subsequent activities and technological innovations, maintaining long-term connections with them.

[0114] By analyzing dissemination data (such as readership, likes, comments, and shares) across different languages, the effectiveness of multilingual information dissemination can be evaluated. Languages ​​and content formats with good dissemination results will be further optimized and promoted. For those with poor results, the reasons will be analyzed and adjustments made. For example, if a particular language version of the information has low readership, it may be due to inaccurate or vivid language expression, or inappropriate platform and timing of publication. These issues will be addressed to improve the international dissemination of the information. Through these methods, using multilingual virtual avatars to publish information can better meet the needs of international participants, ensure effective global dissemination of information, and enhance the international influence and participation of the event.

[0115] Optionally, in step 101, after acquiring the information data to be analyzed, the method further includes: constructing a real-time data stream processing pipeline and using the real-time data stream processing pipeline to process the information data at the millisecond level to extract dynamic analysis indicators corresponding to the hot topic; the dynamic analysis indicators include at least: the growth rate of discussion volume of the hot topic within the time window, sentiment value, keyword frequency, and negative sentiment concentration; inputting the dynamic analysis indicators into the dynamic user profile models of various social media platforms to obtain the group behavior patterns corresponding to the hot topic; comparing the real-time collected dynamic analysis indicators with the dynamic thresholds in the group behavior patterns to identify abnormal risk events corresponding to the hot topic; and immediately activating an emergency response mechanism based on the abnormal risk events to generate and release reassurance information matching the group behavior patterns on various social media platforms to achieve auxiliary positive information guidance for the hot topic.

[0116] In the above embodiments, a real-time data stream processing pipeline is constructed to process information data at the millisecond level, enabling rapid capture of dynamic changes in trending topics. By extracting dynamic analysis indicators such as the growth rate of discussion volume, sentiment value, keyword frequency, and negative sentiment concentration of trending topics within a time window, the development trend of trending topics can be accurately grasped. For example, it can promptly detect sudden increases in topic popularity or changes in public sentiment, providing accurate data support for subsequent information guidance. Furthermore, the dynamic analysis indicators are input into a dynamic user profile model to obtain the group behavior patterns corresponding to trending topics. This helps to gain a deeper understanding of the behavioral characteristics and preferences of different groups in trending topics, such as which groups are more inclined to post negative comments, and which groups are more sensitive to specific keywords. This makes information guidance measures more targeted and improves the guidance effect. Next, comparing the real-time collected dynamic analysis indicators with dynamic thresholds in the group behavior patterns can effectively identify abnormal risk events corresponding to trending topics. Once the indicators exceed the threshold, it means that there may be a potential crisis or risk of negative information outbreak, providing an early warning for timely countermeasures. Finally, by activating emergency response mechanisms based on unusual risk events, and generating and disseminating reassuring messages that match group behavior patterns, it is possible to stabilize public sentiment in a timely manner during crises and guide information in a positive direction. Through targeted information dissemination, public understanding and trust in relevant events can be enhanced, the spread of negative information reduced, and a healthy information environment maintained.

[0117] For example, suppose a well-known sports brand launches a new athletic shoe, sparking widespread discussion on social media. After acquiring relevant data, the real-time data stream processing pipeline rapidly processes the data in milliseconds. Analysis reveals that the discussion volume of this trending topic increased by 200% in the past hour (a growth rate indicator for trending topics within a time window), with a sentiment value of 0.4 (assuming a sentiment value range of -1 to 1, 0.4 indicates a slightly positive sentiment value), a keyword frequency of 50 times per minute (a keyword frequency indicator), and a negative sentiment concentration of 15% (a negative sentiment concentration indicator). These dynamic analysis indicators are input into dynamic user profile models on various social media platforms. Analysis shows that young sports enthusiasts are more concerned with product performance and design, tending to use keywords related to sports technology in their discussions; while fashion enthusiasts focus more on the appearance and styling of shoes, discussing fashion elements more frequently. Furthermore, based on historical data and model analysis, the behavioral patterns and dynamic thresholds of different groups under similar trending topics were determined.

[0118] Continuing to monitor dynamic analysis indicators in real time, it was discovered that the concentration of negative sentiment suddenly rose to 40% two hours later, exceeding the previously set dynamic threshold of 30%. Simultaneously, while the growth rate of discussion volume continued to rise, the rate of increase slowed, potentially indicating some product issues leading to increased public dissatisfaction, thus being identified as an abnormal risk event. Based on this abnormal risk event, the emergency response mechanism was immediately activated. The brand quickly generated and released reassuring messages matching the group behavior patterns on various social media platforms. For young sports enthusiasts, detailed test reports on product performance and user reviews from professional athletes were released, emphasizing the product's sports technology advantages. For fashion enthusiasts, fashion styling suggestions for the new sneakers and trend analysis were released. Through these targeted reassurance messages, public sentiment gradually stabilized, the concentration of negative sentiment began to decrease, and the information gradually shifted in a positive direction, achieving auxiliary positive information guidance for the hot topic.

[0119] In this application's technical solution, distributed data acquisition, real-time analysis models, and automated strategy generation are used to quickly perceive and respond to information changes. This application leverages in-depth analysis through multi-technology integration and targeted strategy formulation to accurately grasp information, enhance guidance effectiveness, and flexibly adjust strategies based on different platform characteristics and dynamic information. This enhances adaptability to complex information environments and unexpected events, achieving full automation and intelligence in the information processing process, simplifying information processing workflows, and improving processing efficiency and quality.

[0120] In another embodiment of this application, an information data processing apparatus is also provided. See also Figure 2 The device comprises the following units:

[0121] The acquisition unit is configured to acquire information data to be analyzed; the information data comes from different social media platforms.

[0122] The identification unit is configured to use a hot topic detection model to identify hot topics in the information data in real time, and track the changing trends of hot topics in the dynamic graph of social networks to obtain information analysis results of hot topics; the information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies;

[0123] The generation unit is configured to generate real-time interaction strategies corresponding to trending topics based on the information analysis results and the types of various social media platforms. The real-time interaction strategies include at least: opinion leader interaction strategies adapted to activity information and social relationship influence, and topic information release strategies adapted to the characteristics of information changes.

[0124] The interaction unit is configured to execute the real-time interaction strategy through a real-time interaction model on various social media platforms to guide positive information about trending topics. The real-time interaction model is used to simulate the behavioral characteristics of users on various social media platforms to obtain information to be published or operation instructions, and push them to users for confirmation of whether to execute.

[0125] Further optionally, after acquiring the information data to be analyzed, the acquisition unit is further configured to:

[0126] The process involves: acquiring user task requirements for the positive information guidance process; constructing a guidance decision model adapted to trending topics in the information data using a big oracle model based on the task requirements and the information data; the guidance decision model including a behavior suggestion module and a long-term monitoring module; using the behavior suggestion module to predict the changing trends of the task requirements and the information data to obtain monitoring behavior suggestions for trending topics; using the long-term monitoring module to match the monitoring methods of the task requirements, the information data, and the monitoring behavior suggestions to obtain a long-term monitoring strategy for trending topics; executing the long-term monitoring strategy on various social media platforms and monitoring the changes in information data after executing the real-time interaction strategy on each social media platform, dynamically adjusting the information release strategy and behavior operation strategy in the real-time interaction strategy.

[0127] Further optionally, the acquisition unit is configured to execute the long-term monitoring strategy on each social media platform as follows:

[0128] Based on the aforementioned long-term monitoring strategy, a long-term information evaluation system matching trending topics is constructed. Using this system, information change predictions are made based on long-term information monitoring data collected from various social media platforms and the information data changes after implementing the real-time interaction strategy, yielding information change index data. A long-term social network dynamic graph is constructed from the long-term information monitoring data, the information change index data, and the information data changes. The degree centrality and betweenness centrality of nodes in the dynamic graph are calculated to assess the social relationship influence of opinion leaders, thus obtaining dynamic information change data corresponding to the long-term social network. Based on the information change index data and the dynamic information change data, a long-term information evaluation report for trending topics is generated.

[0129] Further optionally, the monitoring unit monitors the changes in information data on various social media platforms after the real-time interaction strategy is executed, and dynamically adjusts the information publishing strategy and behavioral operation strategy in the real-time interaction strategy, and is configured as follows:

[0130] Extract information change data after the execution of the real-time interaction strategy from various social media platforms; determine the speed and scope of information dissemination trends on each social media platform based on the information change data; and dynamically adjust the frequency and scope of information release for hot topics according to the speed and scope of information dissemination trends.

[0131] Further optionally, the identification unit employs a hot topic detection model to identify hot topics in the information data in real time, and tracks the changing trends of hot topics in the social network dynamic graph to obtain information analysis results of hot topics, which is configured as follows:

[0132] A trending topic detection model is used to perform semantic analysis on the information data to extract event keywords, event themes, and sentiment trends. Based on the extracted event keywords, event themes, and sentiment trends, user nodes, topic nodes, and content nodes are constructed, and these nodes are connected to form a dynamic social network graph. Event identification is performed based on the dynamic social network graph to obtain the trending topics to be monitored. Trending topics are determined by the access popularity of event nodes, the number of associated edges, and the relevant information density in the dynamic social network graph. Opinion leader activity analysis is performed based on the user nodes in the dynamic social network graph to obtain key opinion leaders related to the trending topics and their influence on the topics. The system analyzes relevant activity information and social relationship influence; it uses graph neural networks (GNNs) to analyze the degree centrality, betweenness centrality, and PageRank value of each user node to identify key opinion leaders; it extracts user nodes, topic nodes, and content nodes related to hot topics from the social network dynamic graph to construct a local social network dynamic graph corresponding to the hot topic; it selects a topic analysis model matching the hot topic from a candidate data model library; and it uses the topic analysis model to perform adaptive topic analysis on the local social network dynamic graph to obtain the information analysis results; wherein, the topic analysis model is pre-trained using historical information data of historical hot topics matching the hot topic.

[0133] Optionally, the system further includes a display unit configured to: after the identification unit uses a hot topic detection model to identify hot topics in the information data in real time and track the changing trends of hot topics in the dynamic graph of the social network, and obtain the information analysis results of the hot topics, extract a set of hot topic indicators from the information analysis results; the set of hot topic indicators includes at least: hot topic evolution records, related keywords, public sentiment analysis results, and opinion leader activity data; the public sentiment analysis results include at least: the proportion of each public sentiment category and the corresponding changing trends; project the set of hot topic indicators onto the visualization space and convert them into dynamic data graphs, and bind the dynamic data graphs to the visualization chart component; the dynamic data graphs include at least: hot topic evolution graph, public sentiment distribution graph, and opinion leader activity distribution graph; generate a hot topic information data analysis report based on the set of hot topic indicators, and display the information data analysis report and the visualization chart component to the user.

[0134] Further optionally, the generation unit, based on the information analysis results and the type of each social media platform, generates a real-time interaction strategy corresponding to the trending topics, configured as follows:

[0135] A sentiment analysis model is used to identify the stance of key opinion leaders based on activity information in the information analysis results, thereby obtaining the stance type of the key opinion leaders. The stance type is one of positive support, neutral observation, or negative skepticism. A multi-dimensional influence assessment model is used to conduct multi-dimensional analysis based on the social relationship influence in the information analysis results to obtain the leader influence level of the key opinion leaders. Based on the leader influence level and the stance type, an opinion leader interaction strategy corresponding to the key opinion leader is generated. The frequency of dynamic strategy adjustment is set according to the event popularity of hot topics, and the leader influence level and the stance type are updated based on the dynamic strategy adjustment frequency. If the leader influence level and the stance type change, the opinion leader interaction strategy is switched through the rule engine bound to the social media platform. The topic content is predicted and the topic construction method is analyzed based on the characteristics of information change to obtain the topic information release strategy. The opinion leader interaction strategy and the topic information release strategy are fused to obtain the real-time interaction strategy.

[0136] Optionally, the system further includes a construction unit configured to: after the acquisition unit acquires the information data to be analyzed, construct a real-time data stream processing pipeline, and use the real-time data stream processing pipeline to process the information data at the millisecond level to extract dynamic analysis indicators corresponding to the hot topic; the dynamic analysis indicators include at least: the growth rate of discussion volume of the hot topic within the time window, sentiment value, keyword frequency, and negative sentiment concentration; input the dynamic analysis indicators into the dynamic user profile models of various social media platforms to obtain the group behavior patterns corresponding to the hot topic; compare the real-time collected dynamic analysis indicators with the dynamic thresholds in the group behavior patterns to identify abnormal risk events corresponding to the hot topic; and immediately activate an emergency response mechanism based on the abnormal risk events to generate and release reassurance information that matches the group behavior patterns on various social media platforms, so as to achieve auxiliary positive information guidance for the hot topic.

[0137] The device can implement various steps in the above method embodiments, which will not be elaborated here.

[0138] In this embodiment, an information data processing device is used, which can quickly sense and respond to information changes, flexibly adjust strategies according to the characteristics of different platforms and dynamic information, enhance the adaptability to complex information environments and emergencies, realize the automation and intelligence of the entire information processing process, simplify the information processing process, and improve processing efficiency and quality.

[0139] Please see Figure 3 , Figure 3 A schematic diagram illustrating an embodiment of the electronic device provided in this application. For example... Figure 3As shown, this application provides an electronic device 500, including a memory 510, a processor 520, and a computer program 511 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 511, it implements an information data processing method.

[0140] Please see Figure 4 , Figure 4 This is a schematic diagram illustrating an embodiment of a computer-readable storage medium provided in this application. For example... Figure 4 As shown, this embodiment provides a computer-readable storage medium 600, on which a software program 611 is stored. When the software program 611 is executed by a processor, it implements an information data processing method.

[0141] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0142] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0144] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An information data processing method, characterized in that, The method includes: Acquire information data to be analyzed; the information data comes from different social media platforms; A hot topic detection model is used to identify hot topics in the information data in real time and track the changing trends of hot topics in the dynamic graph of social networks to obtain information analysis results of hot topics; the information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies; Based on the information analysis results and the types of various social media platforms, real-time interaction strategies corresponding to trending topics are generated, including: using a sentiment analysis model to identify leader positions based on activity information in the information analysis results, to obtain the position type of key opinion leaders; the position type is one of positive support, neutral observation, or negative skepticism; using a multi-dimensional influence assessment model to conduct multi-dimensional analysis based on the social relationship influence in the information analysis results, to obtain the leader influence level of key opinion leaders; generating opinion leader interaction strategies corresponding to key opinion leaders based on the leader influence level and the position type; and setting dynamics according to the event popularity of trending topics. The system adjusts the frequency of dynamic strategy adjustments and updates the leader influence level and the stance type based on the frequency of dynamic strategy adjustments. If the leader influence level and the stance type change, the opinion leader interaction strategy is switched through the rule engine bound to the social media platform. The system predicts topic content and analyzes topic construction methods based on the characteristics of information changes to obtain the topic information release strategy. The system merges the opinion leader interaction strategy and the topic information release strategy to obtain the real-time interaction strategy. The real-time interaction strategy includes at least: an opinion leader interaction strategy adapted to activity information and social relationship influence, and a topic information release strategy adapted to the characteristics of information changes. The real-time interaction strategy is executed through a real-time interaction model to simulate the behavioral characteristics of users on various social media platforms to obtain matching information to be published and / or behavioral operation instructions, and push them to users for confirmation of whether to execute, thereby achieving positive information guidance on trending topics.

2. The information data processing method according to claim 1, characterized in that, After acquiring the information data to be analyzed, the process also includes: Obtain user requirements for the positive information guidance process; Based on the task requirements and the information data, a guidance decision-making model adapted to the hot topics in the information data is constructed using a large language model; the guidance decision-making model includes a behavior suggestion module and a long-term monitoring module; The behavior suggestion module predicts the changing trends of the task requirements and the information data to obtain monitoring behavior suggestions for hot topics. The long-term monitoring module matches the task requirements, information data, and monitoring behavior suggestions to obtain a long-term monitoring strategy for hot topics. The long-term monitoring strategy is implemented on each social media platform, and the changes in information data after the implementation of the real-time interaction strategy on each social media platform are monitored. The information release strategy and behavior operation strategy in the real-time interaction strategy are dynamically adjusted.

3. The information data processing method according to claim 2, characterized in that, The implementation of the long-term monitoring strategy on various social media platforms includes: Based on the aforementioned long-term monitoring strategy, a long-term information evaluation system matching hot topics is constructed; Using the aforementioned long-term information evaluation system, information change prediction is performed on the long-term information monitoring data collected from various social media platforms and the information data changes after the implementation of the aforementioned real-time interaction strategy, thereby obtaining information change index data. A long-term social network dynamic graph is constructed from the long-term information monitoring data, the information change index data, and the information data change situation. The degree centrality and betweenness centrality of the nodes in the social network dynamic graph are calculated to evaluate the social relationship influence of opinion leaders, so as to obtain the dynamic information change data corresponding to the long-term social network. Based on the aforementioned information change index data and the aforementioned dynamic information change data, a long-term information evaluation report on hot topics is generated.

4. The information data processing method according to claim 2, characterized in that, The monitoring of information data changes after the execution of the real-time interaction strategy on various social media platforms, and the dynamic adjustment of information publishing strategies and behavioral operation strategies in the real-time interaction strategy, include: Extract information change data after the execution of the real-time interaction strategy from various social media platforms; Based on the aforementioned information change data, determine the speed and scope of changes in information dissemination trends across various social media platforms; Based on the changing speed and scope of information dissemination trends, dynamically adjust the frequency and scope of information releases on trending topics.

5. The information data processing method according to claim 1, characterized in that, The method employs a trending topic detection model to identify trending topics in the information data in real time and tracks the changing trends of these topics in the dynamic graph of the social network, thereby obtaining information analysis results for the trending topics, including: A hot topic detection model is used to perform semantic analysis on the information data in order to extract event keywords, event themes, and sentiment tendencies from the information data. Based on the extracted event keywords, event themes, and sentiment, user nodes, topic nodes, and content nodes are constructed, and the nodes are connected to form the dynamic graph of the social network. Event identification is performed based on the dynamic graph of the social network to obtain the hot topics to be monitored; hot topics are determined by the access popularity of event nodes, the number of associated edges, and the relevant information density in the dynamic graph of the social network. Based on the user nodes in the aforementioned social network dynamic graph, opinion leader activity analysis is performed to obtain key opinion leaders related to hot topics, their corresponding activity information, and social relationship influence; the degree centrality, betweenness centrality, and PageRank value of each user node are analyzed using a graph neural network (GNN) to identify the key opinion leaders. Extract user nodes, topic nodes, and content nodes related to trending topics from the aforementioned social network dynamic graph, and construct a local social network dynamic graph corresponding to the trending topic. Select a topic analysis model from the candidate data model library that matches the trending topics; The topic analysis model is used to perform adaptive topic analysis on the local social network dynamic graph to obtain the information analysis results. The topic analysis model is pre-trained using historical information data of historical hot topics that are matched with trending topics.

6. The information data processing method according to claim 1, characterized in that, The process of using a trending topic detection model to identify trending topics in the information data in real time, and tracking the changing trends of trending topics in the social network dynamic graph to obtain the information analysis results of the trending topics, further includes: A set of hot topic indicators is extracted from the information analysis results; the set of hot topic indicators includes at least: records of the evolution of hot topics, related keywords, public sentiment analysis results, and data on the activities of opinion leaders; the public sentiment analysis results include at least: the proportion of each public sentiment category and its corresponding trend. The set of hot topic indicators is projected onto the visualization space and converted into dynamic data charts, which are then bound to the visualization chart component. The dynamic data charts include at least: a hot topic evolution chart, a public sentiment distribution chart, and an opinion leader activity distribution chart. Based on the set of hot topic indicators, an information data analysis report on hot topics is generated, and the information data analysis report and the visualization chart component are displayed to the user.

7. The information data processing method according to claim 1, characterized in that, After acquiring the information data to be analyzed, the process also includes: A real-time data stream processing pipeline is constructed, and the information data is processed at the millisecond level using the real-time data stream processing pipeline to extract dynamic analysis indicators corresponding to hot topics; the dynamic analysis indicators include at least: the growth rate of discussion volume of hot topics within the time window, sentiment value, keyword frequency, and negative sentiment concentration; The dynamic analysis indicators are input into the dynamic user profile models of various social media platforms to obtain the group behavior patterns corresponding to trending topics. The real-time collected dynamic analysis indicators are compared with the dynamic thresholds in the group behavior pattern to identify abnormal risk events corresponding to hot topics. Based on the aforementioned abnormal risk events, an emergency response mechanism will be immediately activated to generate and release reassuring messages that match the group behavior patterns on various social media platforms, in order to provide supplementary positive information guidance for trending topics.

8. An information data processing device, characterized in that, The device includes the following units, wherein, The acquisition unit is configured to acquire information data to be analyzed; the information data comes from different social media platforms. The identification unit is configured to use a hot topic detection model to identify hot topics in the information data in real time, and track the changing trends of hot topics in the dynamic graph of social networks to obtain information analysis results of hot topics; the information analysis results include at least: the information change characteristics of hot topics on various social media platforms, the activity information of opinion leaders, and the influence of social relationships; the hot topic detection model is obtained by combining graph neural networks and deep learning; the information change characteristics include at least: information dissemination trends, public sentiment, and emotional tendencies; The generation unit is configured to generate real-time interaction strategies corresponding to trending topics based on the information analysis results and the types of various social media platforms. The real-time interaction strategies include at least: opinion leader interaction strategies adapted to activity information and social relationship influence, and topic information release strategies adapted to the characteristics of information changes. The generation unit, based on the information analysis results and the types of various social media platforms, generates real-time interaction strategies corresponding to trending topics. Specifically, it is configured to: employ a sentiment analysis model to identify leader positions based on activity information in the information analysis results, thereby obtaining the position type of key opinion leaders; the position type is one of positive support, neutral observation, or negative skepticism; use a multi-dimensional influence assessment model to perform multi-dimensional analysis based on social relationship influence in the information analysis results, thereby obtaining the leader influence level of key opinion leaders; generate opinion leader interaction strategies corresponding to key opinion leaders based on the leader influence level and the position type; set a dynamic strategy adjustment frequency according to the event popularity of the trending topic, and update the leader influence level and the position type based on the dynamic strategy adjustment frequency; if the leader influence level and the position type change, switch the opinion leader interaction strategy through the rule engine bound to the social media platform; predict topic content and analyze topic construction methods based on information change characteristics to obtain the topic information release strategy; and fuse the opinion leader interaction strategy and the topic information release strategy to obtain the real-time interaction strategy. The interaction unit is configured to execute the real-time interaction strategy through a real-time interaction model on various social media platforms to guide positive information about trending topics. The real-time interaction model is used to simulate the behavioral characteristics of users on various social media platforms to obtain information to be published or operation instructions, and push them to users for confirmation of whether to execute.

9. An electronic device, characterized in that, Includes memory used to store computer software programs; A processor is configured to read and execute the computer software program, thereby implementing the information data processing method according to any one of claims 1-7.

Images