A social media data aggregation analysis system and method

Abstract

The present invention proposes a social media data aggregation analysis system and method, the method comprising: obtaining social media platform historical data, establishing internal associations with platform unique identifiers and time tags; extracting historical feature data and event data; generating platform portraits based on feature data; generating event portraits based on event data; establishing a data acquisition model based on platform portraits; obtaining event portrait groups through event portrait clustering; generating a cluster analysis model for each event portrait group; obtaining first social data based on the acquisition model; identifying first event data; matching the event analysis model based on the attributes of the first event; and analyzing the first event data based on the matched model to obtain an analysis result. Through the scheme of the present invention, social media data can be intelligently, efficiently, and accurately analyzed to achieve perfect management of social media platforms.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a social media data aggregation analysis system and method. Background Art

[0002] Social networks are developing rapidly, and the number of users is growing explosively. Through social network services, people not only engage in social activities, but also use social networks as public media platforms to meet social needs and obtain specific interests. Social media is a tool and platform for people to share opinions, insights, experiences and views with each other. At this stage, it mainly includes social networking sites, Weibo, WeChat, blogs, forums, podcasts, etc. At present, how to accurately and efficiently analyze social media data is still a problem that needs to be solved. Summary of the invention

[0003] Based on the above problems, the present invention proposes a social media data aggregation analysis system and method. Through the scheme of the present invention, social media data can be analyzed intelligently, efficiently and accurately to achieve perfect management of social media platforms.

[0004] In view of this, one aspect of the present invention proposes a social media data aggregation and analysis system, comprising: a cloud server and a social media platform;

[0005] The cloud server is configured as follows:

[0006] Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships;

[0007] Extracting historical feature data and historical event data of the social media platform from the historical data;

[0008] Generate multiple social media platform portrait data based on the historical feature data;

[0009] Generate multiple event portrait data according to the historical event data;

[0010] Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data;

[0011] Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups;

[0012] Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models;

[0013] Acquire first social media data from the social media platform according to the first data acquisition model;

[0014] identifying first event data from the first social media data;

[0015] Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data of the first event data;

[0016] The first event data is analyzed according to the first event data analysis model to obtain a first analysis result.

[0017] Another aspect of the present invention provides a social media data aggregation analysis method, comprising:

[0018] Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships;

[0019] Extracting historical feature data and historical event data of the social media platform from the historical data;

[0020] Generate multiple social media platform portrait data based on the historical feature data;

[0021] Generate multiple event portrait data according to the historical event data;

[0022] Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data;

[0023] Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups;

[0024] Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models;

[0025] Acquire first social media data from the social media platform according to the first data acquisition model;

[0026] identifying first event data from the first social media data;

[0027] Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data of the first event data;

[0028] The first event data is analyzed according to the first event data analysis model to obtain a first analysis result.

[0029] Optionally, the step of obtaining historical data of multiple social media platforms and marking the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships includes:

[0030] Obtain developer documentation of the social media platform, and extract the API interface and access permissions provided by the social media platform;

[0031] Acquire the historical data according to the API interface and access rights of the social media platform;

[0032] When acquiring the historical data, marking the historical data using the unique identifier of the social media platform and the time relationship;

[0033] The historical data is stored in a database, and association modeling is performed on the historical data according to the unique identifier of the social media platform and the time relationship to obtain the association relationship.

[0034] Optionally, the step of generating multiple social media platform portrait data according to the historical feature data includes:

[0035] Determining the types of features included in the social media platform profile data;

[0036] According to the feature type, feature engineering is performed on the historical feature data to obtain first feature data;

[0037] Based on the first characteristic data, a plurality of social media platform portrait data are generated.

[0038] Optionally, the step of generating a plurality of event portrait data according to the historical event data includes:

[0039] Use natural language processing technology to extract event elements from the historical event data to obtain event information including event subject, time, location, cause, development process, and transmission path;

[0040] Applying knowledge graph technology to process the event information and construct an event relationship graph;

[0041] Use sentiment analysis technology and intention recognition technology to analyze the historical event data and determine the emotional tendency data of participants and the public;

[0042] Using a social network analysis algorithm to identify the spread scope and spread model of the event from the historical event data;

[0043] Using multimodal analysis technology to associate text data and multimedia data in the historical event data to build an event knowledge base;

[0044] A structured event portrait is generated according to the event relationship diagram, the emotional tendency data, the propagation scope, the propagation model, and the event knowledge base to obtain the event portrait data.

[0045] Optionally, the step of establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data includes:

[0046] Determine the user portrait, content portrait, interaction portrait, platform positioning and scenario, security policies and restrictions of the social media platform based on the social media platform portrait data;

[0047] The first data acquisition model is determined according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policies and restrictions.

[0048] Optionally, the step of determining the first data acquisition model according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policy and restriction includes:

[0049] According to the user portrait, determine the data collection granularity for different user characteristics;

[0050] Determine the data type / category to be collected based on the content portrait;

[0051] Determine the collected data of core node users based on the interaction portrait;

[0052] According to the platform positioning and scenarios, customize the data collection strategy and determine the media type, time range, and collection frequency to be collected;

[0053] Determine data access rights and authorization mechanisms based on the security policies and restrictions;

[0054] The data acquisition method is determined based on the technical attributes of the social media platform.

[0055] Optionally, the step of acquiring first social media data from the social media platform according to the first data acquisition model includes:

[0056] Parsing the collection requirements, user groups, collection keywords, collection methods, and collection frequency of the social media platform from the first data acquisition model;

[0057] The first social media data is acquired according to the acquisition requirement, the user group, the acquisition keyword, the acquisition method, and the acquisition frequency.

[0058] Optionally, the step of identifying first event data from the first social media data includes:

[0059] Inputting the first social media data into a preset first event recognition model;

[0060] The first event recognition model applies a text similarity algorithm to compare the similarity between the short text extracted from the first social media data and the historical event brief introduction to obtain a candidate event set;

[0061] Using named entity recognition and entity linking technology, identifying event subjects and relationships between event subjects from the first social media data;

[0062] By means of the knowledge graph, the relationship between event subjects is analyzed to determine the validity value of each candidate event in the candidate event set;

[0063] Clustering each of the candidate events, and filtering noise data using statistical characteristics of the clusters;

[0064] Determine the time range and geographical scope of the event by combining the time and location information identified from the first social media data;

[0065] Sorting each of the candidate events according to the validity value, and calculating the confidence score of the event according to each dimensional feature;

[0066] A first candidate event with the highest confidence is selected and output as the first event data.

[0067] Optionally, the method for constructing the first event recognition model includes:

[0068] Constructing a training data set for first event recognition and annotating event elements;

[0069] Use the sequence labeling model to train the event feature extraction model;

[0070] Dictionaries and rule libraries related to the first event are collected, keyword matching and rule matching are performed, and the first event recognition model is obtained by combining the event element extraction model.

[0071] According to the technical solution of the present invention, the social media data aggregation analysis method includes obtaining historical data of multiple social media platforms, and marking the historical data with the unique identifier and time relationship of the social media platform to establish an internal association relationship; extracting historical feature data and historical event data of the social media platform from the historical data; generating multiple social media platform portrait data according to the historical feature data; generating multiple event portrait data according to the historical event data; establishing a first data acquisition model for each social media platform according to the social media platform portrait data; performing cluster analysis according to the event portrait data to obtain multiple event portrait data groups; generating a corresponding data cluster analysis model for each event portrait data group to obtain multiple event data analysis models; obtaining first social media data from the social media platform according to the first data acquisition model; identifying first event data from the first social media data; determining the corresponding first event data analysis model from the multiple event data analysis models according to the attribute data of the first event data; and analyzing the first event data according to the first event data analysis model to obtain a first analysis result. According to the solution of the present invention, social media data can be intelligently, efficiently and accurately analyzed to achieve perfect management of social media platforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a schematic block diagram of a social media data aggregation and analysis system provided by an embodiment of the present invention;

[0073] Figure 2 It is a flow chart of a social media data aggregation analysis method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0074] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0075] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0076] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0077] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0078] Refer to the following Figure 1 to Figure 2 A social media data aggregation analysis system and method provided according to some embodiments of the present invention will be described.

[0079] like Figure 1 As shown, one embodiment of the present invention provides a social media data aggregation and analysis system, including: a cloud server and a social media platform;

[0080] The cloud server is configured as follows:

[0081] Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships;

[0082] Extracting historical feature data and historical event data of the social media platform from the historical data;

[0083] Generate multiple social media platform portrait data based on the historical feature data;

[0084] Generate multiple event portrait data according to the historical event data;

[0085] Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data;

[0086] Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups;

[0087] Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models;

[0088] Acquire first social media data from the social media platform according to the first data acquisition model;

[0089] identifying first event data from the first social media data;

[0090] Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data of the first event data;

[0091] The first event data is analyzed according to the first event data analysis model to obtain a first analysis result.

[0092] It should be known that Figure 1 The block diagram of the social media data aggregation analysis system shown is for illustration only, and the number of modules shown does not limit the protection scope of the present invention. The social media data aggregation analysis system provided in this embodiment can be used to execute the corresponding embodiments of the social media data aggregation analysis method. For the specific implementation process, please refer to the description of the following method embodiments, which will not be repeated here.

[0093] See also Figure 2 Another embodiment of the present invention provides a social media data aggregation analysis method, comprising:

[0094] Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships;

[0095] Extracting historical feature data and historical event data of the social media platform from the historical data;

[0096] Generate multiple social media platform portrait data based on the historical feature data;

[0097] Generate multiple event portrait data according to the historical event data;

[0098] Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data;

[0099] Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups;

[0100] Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models;

[0101] Acquire first social media data from the social media platform according to the first data acquisition model;

[0102] identifying first event data from the first social media data;

[0103] Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data of the first event data;

[0104] The first event data is analyzed according to the first event data analysis model to obtain a first analysis result.

[0105] In an embodiment of the present invention, historical feature data may be attribute data describing users, posts, interactive behaviors, etc., such as user registration time, post publishing time, number of likes, number of comments, etc.; historical event data may be events occurring on specific social media platforms, such as important events, hot topics, user activity trends, etc.

[0106] In the embodiment of the present invention, the historical data includes but is not limited to: user data, including the number of registered users of the platform, user attributes (such as age, gender, geographic location), user activity (such as monthly active users, daily active users), user growth trend, etc.; content data, including the number and type of posts, pictures, videos, comments and other content posted by users, which can be used to analyze user behavior, content trends and interests; interactive data, including interactive behaviors between users, such as likes, comments, shares, forwarding, etc., which can reflect the user's reaction to and participation in the content; advertising data, including the number of advertisements, the number of advertisers, the click-through rate of advertisements, the revenue from advertisements, etc.; social relationship data, including the attention relationship between users. Social relationship data such as relationships, friendships, and fan relationships can be used to analyze user social networks and influence; group data, that is, data on various groups or communities on social platforms, including group members, group management, group activities, etc., which can be used to analyze group structure and behavior; feedback data, including user feedback, reports, complaints, etc. on the platform, which can help the platform improve user experience and solve problems; security data, including data on security incidents and violations on the platform, which can be used to monitor and respond to violations and maintain the security and order of the platform; analytical data, including analysis results and statistical data on the above data, which can be used to generate reports, trend analysis, and decision support, etc. Obtaining historical data of these different dimensions can comprehensively analyze and model a social platform.

[0107] Event profile data includes but is not limited to:

[0108] Origin of the event: Describe the origin and triggering factors of the event. This can be a specific event, behavior, speech, photo, video or other content that attracts attention; Key participants in the event: Identify and describe the key participants in the event, including individuals, organizations, communities, media, etc. These participants can be the initiators, promoters, supporters, opponents or those affected by the event; Time and place of the event: Record the time and place of the event, which helps to track the timeline and geographical scope of the event in order to understand the evolution and scope of the event; Content of the event: Describe the specific content and details of the event. This includes related posts, comments, reposts, videos, pictures and other media content, as well as event-related tags, keywords, etc.; Event sentiment and attitude: Analyze the sentiment and attitude of event participants and social media users towards the event. This can be done through sentiment analysis, comment analysis and other methods to understand the user's reaction and emotional tendency to the event; Event propagation and impact: Describe the propagation path, scope of influence and propagation speed of the event on the social media platform, which includes the interaction between users, the path of information propagation, key nodes and influencers, as well as the popularity and influence of the event on social media; Feedback and response: Record the feedback and response of social media users to the event. This can be comments, replies, likes, shares and other interactive behaviors, which help to understand users' views on and participation in the event; Media coverage and public opinion influence: record media coverage of the event and public opinion influence. This includes the views and comments of news media, columnists, opinion leaders, etc., and the degree of influence of media reports on the spread and impact of the event; Event trend and subsequent development: predict and analyze the trend and subsequent development of the event based on the evolution of the event and related factors; and so on.

[0109] According to the technical solution of this embodiment, the social media data aggregation analysis method includes obtaining historical data of multiple social media platforms, and marking the historical data with the unique identifier and time relationship of the social media platform to establish an internal association relationship; extracting historical feature data and historical event data of the social media platform from the historical data; generating multiple social media platform portrait data according to the historical feature data; generating multiple event portrait data according to the historical event data; establishing a first data acquisition model for each social media platform according to the social media platform portrait data; performing cluster analysis according to the event portrait data to obtain multiple event portrait data groups; generating a corresponding data cluster analysis model for each event portrait data group to obtain multiple event data analysis models; obtaining first social media data from the social media platform according to the first data acquisition model; identifying first event data from the first social media data; determining the corresponding first event data analysis model from the multiple event data analysis models according to the attribute data of the first event data; and analyzing the first event data according to the first event data analysis model to obtain a first analysis result. Through the solution of the present invention, social media data can be intelligently, efficiently and accurately analyzed to achieve perfect management of social media platforms.

[0110] In some possible implementations of the present invention, the step of obtaining historical data of multiple social media platforms and marking the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships includes:

[0111] Obtain developer documentation of the social media platform, and extract the API interface and access permissions provided by the social media platform;

[0112] Acquire the historical data according to the API interface and access rights of the social media platform (user information, posts, comments, interactive behaviors, etc. may be acquired according to the functions provided by the API);

[0113] When acquiring the historical data, marking the historical data using a unique identifier of the social media platform (e.g., user ID, post ID) and a time relationship (e.g., posting time);

[0114] Storing the historical data in a database (a relational database (such as MySQL) or a non-relational database (such as MongoDB) may be used for storage, ensuring that the structure and association of the data are correctly established and maintained);

[0115] According to the unique identifier and time relationship of the social media platform, the historical data is associated with modeling to obtain the association relationship (data association algorithms and techniques, such as association analysis, graph analysis, time series analysis, etc., can be used to identify and establish internal association relationships between data).

[0116] In this embodiment, by establishing internal associations between data, high efficiency and accuracy of subsequent data processing and profiling can be guaranteed.

[0117] In some possible implementations of the present invention, the step of generating multiple social media platform portrait data according to the historical feature data includes:

[0118] Determine the types of features included in the social media platform portrait data (these features may include platform positioning, functions, user groups, content classification, user attributes (such as age, gender, region), interests, activity, social relationships, etc.);

[0119] According to the feature type, feature engineering is performed on the historical feature data (features may be processed and converted using statistical methods, machine learning algorithms, or domain expert knowledge; for example, a user's activity index, a user's interest classification, a user's social network relationship, etc. may be calculated) to obtain first feature data;

[0120] Based on the first characteristic data, a plurality of social media platform portrait data are generated.

[0121] In the embodiments of the invention, multi-dimensional historical data such as user features, content features, and interaction features from different modules can be integrated, and machine learning algorithms such as clustering and topic models can be used to segment users to obtain a user group model; natural language processing can be used to perform content topic modeling analysis to obtain a content portrait of the platform; a network analysis algorithm can be used to analyze user social relationships and interaction models to find influential nodes; the extracted multi-dimensional feature data can be visualized, such as geographic distribution, age distribution, network topology, etc.; a platform knowledge base based on a knowledge graph can be constructed to store structural knowledge such as entities, attributes, and relationships of the platform; a content generation model based on platform type and positioning can be trained to simulate platform-characterized content; transfer learning technology can be used to fine-tune a pre-trained language model to obtain a platform-specific text representation; sample matching and evaluation can be performed in combination with user survey data to correct the generation results of the platform portrait; feature extraction, data annotation, and model training can be continuously iterated to optimize to make the platform portrait results more accurate and realistic; a variety of artificial intelligence technologies can be used to comprehensively describe multi-dimensional features such as platform users, content, and interactions, and to construct a high-quality social media platform portrait to provide data support for subsequent applications.

[0122] In this embodiment, by using a variety of analysis techniques to construct a platform portrait, the characteristics of the social platform can be fully understood from multiple angles, providing support for subsequent user behavior prediction.

[0123] In some possible implementations of the present invention, the step of generating a plurality of event portrait data according to the historical event data includes:

[0124] Use natural language processing technology to extract event elements from the historical event data to obtain event information including the event subject, time, location, cause, development process, and propagation path (use natural language processing technology to process text data to identify and extract key events, topics, or themes. For example, text classification, topic modeling, named entity recognition, and other technologies can be used to identify and extract important events or topics);

[0125] Apply knowledge graph technology to process the event information (such as linking the event subject and related entities) and construct an event relationship graph;

[0126] Use sentiment analysis technology and intention recognition technology to analyze the historical event data and determine the emotional tendency data of participants and the public;

[0127] Using a social network analysis algorithm to identify the spread scope and spread model of the event from the historical event data;

[0128] Using multimodal analysis technology to associate text data and multimedia data in the historical event data to build an event knowledge base;

[0129] A structured event portrait is generated according to the event relationship diagram, the emotional tendency data, the propagation scope, the propagation model, and the event knowledge base to obtain the event portrait data.

[0130] In this embodiment, it also includes: using transfer learning technology to train the vector representation of the event text in the historical event data for event similarity analysis, dividing the events that reach a preset similarity threshold into the same group, and obtaining multiple event groups; performing the above-mentioned processing on the historical data of each event in the event group, and obtaining portrait data of the multiple event groups.

[0131] In this embodiment, a variety of artificial intelligence technologies are used in combination to describe the intrinsic attributes, external relations, and evolution laws of events from multiple dimensions, which can effectively construct an intelligent portrait of the event and provide decision support for subsequent applications.

[0132] In some possible implementations of the present invention, the step of establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data includes:

[0133] Determine the user portrait, content portrait, interaction portrait, platform positioning and scenario, security policies and restrictions of the social media platform based on the social media platform portrait data;

[0134] In this step, determining the user profile, content profile, interaction profile, platform positioning and scenarios, security policies and restrictions of the social media platform based on the social media platform profile data means determining and describing the user characteristics, content characteristics, interaction characteristics, positioning and usage scenarios, and security policies and restrictions of the platform based on the data of the social media platform.

[0135] User portrait: It can refer to the description and classification of users of social media platforms. By analyzing the user's personal information, interests, behavior patterns, social relationships and other data, a user portrait can be formed to understand the user's characteristics, preferences and behavior habits.

[0136] Content portrait: It can refer to the description and classification of content on social media platforms. By analyzing the published text, pictures, videos and other content information, we can understand the theme, emotional tendency, trend, etc. of the content and form a content portrait to better understand and recommend related content.

[0137] Interaction portrait: It refers to the description and analysis of user interaction behaviors on social media platforms. By analyzing the interaction behaviors between users, such as following, liking, commenting, and sharing, we can understand the relationship between users, social influence, interaction patterns, etc., and form an interaction portrait to better understand users' social behaviors and social network structure.

[0138] Platform positioning and scenarios: This refers to the description of the positioning and usage scenarios of social media platforms. By analyzing the platform's functions, characteristics, user groups, etc., we can understand the platform's positioning in social, entertainment, business, etc., as well as the user's usage habits and behavior patterns in different scenarios.

[0139] Security policies and restrictions: can refer to the policies and restrictions that social media platforms take to protect user safety and maintain order on the platform. This may include account verification, content review, privacy protection, user reporting mechanisms, etc. to ensure user safety and a good user experience on the platform.

[0140] By analyzing and understanding the portrait data of the social media platform, it can help the platform operators understand user characteristics, content trends, interactive behaviors, and the platform's positioning and security strategies, so as to better perform user management, content recommendation, platform operation and security protection.

[0141] The first data acquisition model is determined according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policies and restrictions.

[0142] In this step, based on information such as user portrait, content portrait, interaction portrait, platform positioning and scenario, security policies and restrictions, etc., determining the first data acquisition model may include the following aspects:

[0143] Target user screening: Determine the target user group of the first data acquisition model based on user portraits and platform positioning in order to obtain relevant data in a targeted manner.

[0144] Content screening and recommendation: Based on the content portrait and user portrait, determine the content category that the first data acquisition model focuses on, and recommend relevant content based on user interests in order to obtain data related to the target user.

[0145] Interaction behavior modeling: Based on the interaction portrait and user portrait, the interaction behavior between users is modeled to understand the relationship, influence and interaction pattern between users, so as to obtain data related to user interaction.

[0146] Security and compliance considerations: According to the security policies and restrictions of the platform, ensure that the operation of the first data acquisition model complies with the rules of the platform to ensure the legality and compliance of the data acquisition process.

[0147] In summary, based on information such as user portraits, content portraits, interaction portraits, platform positioning and scenarios, security policies and restrictions, the target users, content of interest, and interactive behavior modeling methods of the first data acquisition model are determined, and the safety and compliance of the operation are ensured. In this way, the first data related to the target users and target content can be obtained more specifically.

[0148] In some possible implementations of the present invention, the step of determining the first data acquisition model according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policy and restrictions includes:

[0149] According to the user portrait, determine the data collection granularity for different user characteristics;

[0150] In this step, according to the different user characteristics described in the user portrait, the degree of data refinement or detailedness adopted for these characteristics during the data collection process is determined. Specifically, the user portrait is information that describes and classifies the user, usually including the user's personal information, hobbies, behavior patterns, social relationships, etc. Different user characteristics may include age, gender, geographic location, occupation, interest field, consumption behavior, etc. Each feature can provide specific information about the user. When determining the granularity of data collection, it is necessary to consider the importance and influence of different user characteristics. Some characteristics may be very critical to our data collection goals, while other characteristics may have little impact on the goals. Therefore, according to the characteristic information in the user portrait, the degree of data refinement of these characteristics during the data collection process can be determined. For example, for the user characteristic of gender, the data collection granularity can be to collect the user's gender information to understand the behavioral differences and preferences of users of different genders on the platform. For the geographic location feature, the user's geographic location information can be collected to understand the usage habits and regional characteristics of users in different regions. In addition, the data collection granularity can also be determined based on other user characteristics. For the interest field feature, the user's behavior and interaction data in different interest fields can be collected to understand the user's interest in different topics. For consumption behavior characteristics, users’ purchase records and consumption preferences can be collected for personalized recommendations and market analysis. Therefore, determining the data collection granularity for different user characteristics based on user portraits means deciding which features to focus on and refine during data collection based on the user characteristics described in the user portraits, so as to more accurately understand the user’s behavior, needs, and preferences, thereby supporting more accurate data analysis and decision-making.

[0151] Determine the data type / category (such as keywords, topics, etc.) to be collected based on the content portrait;

[0152] In this step, based on the analysis results of the content portrait, determine the specific data types or categories to be collected during the data collection process, such as keywords, topics, etc. Specifically, the content portrait is information that describes and classifies the content on the social media platform. The content portrait may include the theme, emotional tendency, trend, etc. of the content. By analyzing the content portrait, you can understand the characteristics and features of the text, pictures, videos and other content posted by users on the social media platform. When determining the data type or category during the data collection process, you can choose specific content attributes to focus on based on the content portrait. For example:

[0153] Keywords: Based on the keyword analysis results in the content portrait, you can determine the specific keywords that need to be collected in order to understand the user's discussion and attention to certain topics or keywords. In this way, you can collect data related to specific keywords.

[0154] Topic: Based on the topic analysis results in the content portrait, you can determine the specific topics that need to be collected in order to understand users' discussions and opinions on certain topics. This way, you can collect data related to specific topics.

[0155] Emotional tendency: Based on the results of the emotional analysis in the content portrait, we can determine the content with specific emotional tendencies that need to be collected, such as positive, negative, or neutral content. In this way, we can collect data related to specific emotional tendencies.

[0156] In addition to the above examples, the data types or categories that need to be collected can also be determined based on other characteristics of the content portrait. This depends on the results of the content portrait analysis and the specific research or application requirements. Therefore, based on the analysis results of the content portrait, determine the specific data types or categories that are concerned and collected during the data collection process, so as to more accurately understand and analyze the content characteristics and trends that users are concerned about.

[0157] Determine the collected data of core node users based on the interaction portrait;

[0158] In this step, based on the analysis results of the interaction portrait, determine the relevant data of the core node users to be collected during the data collection process. Specifically, the interaction portrait is information that describes and analyzes the interactive relationship between users of the social media platform. The interaction portrait may include the relationship network, interaction frequency, interaction behavior, etc. between users. By analyzing the interaction portrait, you can understand the relationship, influence, and interaction pattern between users. When determining the data of the core node users to be collected during the data collection process, you can select the core users to focus on based on the interaction portrait. Core node users are usually users who have high influence on the social media platform, have close relationships with other users, and play an important role in the interaction. Specific data collection may include the following aspects:

[0159] Focus on the interactive behavior of core node users: Collect interactive behavior data between core node users and other users, such as likes, comments, shares, etc. This can help understand the interactive patterns and influence of core node users.

[0160] Analyze the relationship network of core node users: Collect the relationship data between core node users and other users, such as follow-up relationships, friend relationships, fan relationships, etc. This can help understand the social influence and interpersonal relationships of core node users on social media platforms.

[0161] Collect content publishing data of core node users: Collect content data published by core node users, such as text, pictures, videos, etc. This can help understand the interest areas, content characteristics and influence of core node users.

[0162] By collecting relevant data of core node users, we can further study and analyze the core users’ behaviors, influence, and interactive relationships with other users on social media platforms. This is of great significance to fields such as social media marketing and social network analysis. In this step, based on the analysis results of the interactive portrait, we determine the relevant data of the core node users that we focus on and collect during the data collection process, so as to further understand and analyze the interactive behaviors, relationship networks, and content characteristics of core users on social media platforms.

[0163] According to the platform positioning and scenarios, customize the data collection strategy and determine the media type, time range, and collection frequency to be collected;

[0164] In this step, according to the positioning and application scenarios of the platform, formulate a suitable data collection strategy, determine the media type to be collected, time range and collection frequency. Specifically, platform positioning and scenario: refers to the positioning and usage scenario of the social media platform, such as whether it is a comprehensive social platform for mass users or a social platform focusing on a specific field or industry. Different platform positioning and scenarios determine the differences in user groups, content types and usage habits. Data collection strategy: formulate a specific strategy for data collection according to the positioning and scenario of the platform. This includes determining the media type to be collected, time range and collection frequency. Media type: refers to the content media type to be collected, including text, pictures, videos, etc. According to the characteristics of the platform and application requirements, determine the media type to be collected in order to obtain comprehensive data. Time range: determine the time range of data collection, that is, when to start collecting data and the duration of data collection. This can be determined according to the needs of research or analysis, such as collecting data for the past year or long-term historical data. Collection frequency: determine the frequency of data collection, that is, the time interval for collecting data. You can choose to collect regularly, such as once a day, week or month, or collect according to specific events or needs. By developing a data collection strategy based on the platform positioning and scenarios, you can ensure that valid data related to the research or analysis objectives are collected, and timely update and adjust according to the characteristics and changes of the data. In this step, according to the positioning and application scenarios of the social media platform, develop a suitable data collection strategy, clarify the content media type to be collected, time range and collection frequency, so as to obtain valid data related to the research or analysis objectives.

[0165] Determine data access rights and authorization mechanisms based on the security policies and restrictions;

[0166] Determine the data acquisition method based on the technical attributes of the social media platform (such as using an API interface, crawling web page data, subscribing to data streams, etc.).

[0167] In this embodiment, the platform portrait is used comprehensively to guide data collection, so that differentiated data of each platform can be obtained in a standardized manner, laying a data foundation for subsequent analysis and modeling.

[0168] It can be understood that, in this embodiment, the first data acquisition model is used to regulate and guide the collection of data from different social media platforms, including but not limited to the following contents:

[0169] Platform identification: record basic identification information such as the platform's name, type, and main functions.

[0170] Data requirements: Define the data type, time span, and collection volume requirements that need to be collected on the platform based on the analysis objectives.

[0171] User group model: describes the characteristics of the platform's target user group, including demographics, interests, etc.

[0172] Content model: content portraits such as the platform’s common content types, discussion topics, and degree of influence.

[0173] Relationship network model: the social relationship network topology between platform users.

[0174] Data collection API: API information such as the platform's open data interface, authentication mechanism, call restrictions, etc.

[0175] Collection logic: technical implementation solutions such as collection process, incremental crawling logic, etc. designed for the platform.

[0176] Data filtering rules: rules for processing and filtering collected data, such as crawler trap filtering.

[0177] Data storage strategy: strategy and data model for classifying and storing collected data.

[0178] Collection evaluation mechanism: evaluate collection coverage, data quality, efficiency and other indicators, and optimize collection strategies.

[0179] In this embodiment, by defining these contents, guidance and specifications can be provided for the collection work of different platforms, making data acquisition more targeted and systematic.

[0180] In some possible implementations of the present invention, the step of acquiring the first social media data from the social media platform according to the first data acquisition model includes:

[0181] Parsing the collection requirements, user groups, collection keywords, collection methods, and collection frequency of the social media platform from the first data acquisition model;

[0182] The first social media data is acquired according to the acquisition requirement, the user group, the acquisition keyword, the acquisition method, and the acquisition frequency.

[0183] Specifically, call the platform's open API, pass in the access token, and extract the specified user or topic data. Set the crawling time range, number limit, etc.; study the platform's front-end structure, use the crawler program to simulate login and browsing, and crawl the data of specific web page elements; filter and deduplicate the crawled data, extract structured fields such as text, tags, and geographic location, and store them in the database; according to the incremental crawling logic, record the crawled data indicators to avoid repeated crawling; execute sentiment analysis and other algorithms to extract derived feature information such as text sentiment tendency and event elements; according to the platform relationship network model, diffuse and crawl the data of related users to expand the data range; based on the collection and evaluation mechanism, count the amount of crawled data, evaluate the quality, and feedback to optimize the data acquisition model; continuously iterate the crawler strategy, simulate user behavior to bypass anti-crawling detection, and obtain more high-quality data; combine the cloud computing platform to realize distributed crawlers and automated crawling task pipelines.

[0184] In this embodiment, the comprehensive use of various dimensional indicators of the data acquisition model can effectively guide the data capture of the social platform and obtain first-hand data that meets the analysis objectives.

[0185] In some possible implementations of the present invention, the step of identifying the first event data from the first social media data includes:

[0186] Inputting the first social media data into a preset first event recognition model;

[0187] The first event recognition model applies a text similarity algorithm to compare the similarity between the short text extracted from the first social media data and the historical event brief introduction to obtain a candidate event set;

[0188] In this step, the first event recognition model applies a text similarity algorithm to compare the similarity between the short text extracted from the first social media data and the historical event profile, and the candidate event set is obtained. This means that for the event recognition task, the short text extracted from the social media data is compared with the historical event profile using a text similarity algorithm to determine the similarity between them, and a set of candidate events is screened out according to the similarity. In this process, there is first an event recognition model, the purpose of which is to identify content related to a specific event from social media data. This model can be a model based on machine learning or deep learning, which is used to automatically identify and extract information related to events. In order to determine candidate events related to historical events, the model applies a text similarity algorithm, which is a method for comparing the degree of similarity between texts. The text similarity algorithm can measure the semantic or structural similarity between two texts. In this case, the model compares the short text extracted from the social media data with the historical event profile. The historical event profile is a pre-defined and stored brief description or summary related to each event. By calculating the similarity between the short text and the historical event profile, the model can determine the degree of relevance between them. Short texts with higher similarity will be considered to be related to a historical event, thereby forming a candidate event set. The candidate event set is a set of events that may be related to historical events that are screened out after text similarity comparison. These events may require further verification and analysis to determine whether they actually correspond to historical events. By applying a text similarity algorithm and comparing the similarity of short texts with historical event profiles, it can help automatically identify and screen potential events related to historical events and improve the accuracy and efficiency of event recognition.

[0189] Using named entity recognition and entity linking technology, identifying event subjects and relationships between event subjects from the first social media data;

[0190] In this step, using named entity recognition and entity linking technology to identify event subjects and relationships between event subjects from the first social media data means using named entity recognition and entity linking technology to analyze and identify event subjects (Entities) in social media data and the relationships between them. Named Entity Recognition (NER) is a natural language processing technology used to identify entities with specific names in text, such as names of people, places, names of organizations, etc. In the context of event recognition, named entity recognition can help identify event subjects involved in social media data, such as specific people, places, organizations, etc. Entity Linking is the process of linking identified named entities to entities in a knowledge base or database. Through entity linking technology, the identified event subjects can be associated with related entities in external knowledge resources, such as linking names to character databases, linking place names to geographic location databases, etc. Therefore, using named entity recognition and entity linking technology, the first social media data can be processed and analyzed, the event subjects therein can be identified, and they can be linked to related entities in external knowledge bases. This allows the establishment of relationships between event subjects, such as the association between people and places, the association between organizations and places, etc. By identifying event subjects and establishing relationships between them, we can better understand event information in social media data and reveal details about the participants, places, organizations, etc. This helps further event analysis, association analysis, and inference, thereby providing a deeper and more comprehensive understanding of events.

[0191] By means of the knowledge graph, the relationship between event subjects is analyzed to determine the validity value of each candidate event in the candidate event set;

[0192] In this step, the relationship between event subjects is analyzed with the help of knowledge graph, and judging the validity value of each candidate event in the candidate event set means using knowledge graph to analyze the relationship between event subjects and evaluate the validity of each candidate event in the candidate event set. Knowledge graph is a structured knowledge representation method that presents the relationship between entities in the form of a graph. It contains rich knowledge and semantic associations, which can be used to describe the attributes, categories and associations of entities. In this case, with the help of knowledge graph, the relationship between event subjects can be explored. By analyzing the entities and relationships in the knowledge graph, the association between event subjects can be understood, such as the relationship between people, the relationship between organizations and places, etc. The validity evaluation of each candidate event in the candidate event set means analyzing and judging these candidate events to determine their value and relevance. By combining the information in the knowledge graph, the validity value of the candidate event can be evaluated. This may involve considering factors such as the closeness of the relationship between event subjects, the authority of event subjects, and the relevance of event subjects to specific fields. The validity value can be used to measure the importance and relevance of candidate events. According to the evaluation results of the validity value, the candidate events can be sorted and filtered to determine the most valuable and relevant events. By using knowledge graphs to analyze event subject relationships and evaluate the effectiveness of candidate events, we can provide deeper insights and judgments to help determine which candidate events are the most meaningful and valuable for a given task or goal.

[0193] Clustering each of the candidate events, and filtering noise data using statistical characteristics of the clusters;

[0194] In this step, clustering each of the candidate events and filtering the noise data using the statistical characteristics of the clusters means clustering the candidate events and filtering out the noise data using the statistical characteristics in the clustering results. In this process, the candidate events are first clustered, that is, similar events are classified into the same category or cluster. Clustering is an unsupervised learning method that aims to group data according to their similarity. By clustering the candidate events, similar events can be grouped together to form different event clusters. Each event cluster represents a class of similar events. Next, the statistical characteristics in the clustering results are used to filter out the noise data. The clustering results usually provide statistical properties of each cluster, such as the size, density, average value, etc. of the cluster. By analyzing the statistical characteristics in the clustering results, candidate events with abnormal or noise characteristics can be identified and filtered out. These abnormal events may be irrelevant or low-quality events due to data errors, abnormal conditions, or other reasons. By using the statistical characteristics of clustering, these noise data can be excluded from the candidate event set, thereby improving the quality and accuracy of the event data. Therefore, clustering the candidate events and using the statistical features in the clustering results to filter the noise data can help extract high-quality and highly relevant events, thereby providing more reliable and useful information for subsequent analysis and decision-making.

[0195] Determine the time range and geographical scope of the event by combining the time and location information identified from the first social media data;

[0196] Sorting each of the candidate events according to the validity value, and calculating the confidence score of the event according to each dimensional feature;

[0197] In this step, the candidate events are sorted according to the validity value, and the confidence score of the event is calculated according to each dimension feature. This means that the candidate events are sorted according to the validity value mentioned above, and the confidence score of each event is calculated according to multiple features. First, the candidate events are sorted according to the validity value mentioned above. The validity value is an indicator used to measure the importance and relevance of the candidate events. According to the high and low validity values, the candidate events can be sorted, and the most valuable and relevant events are placed in front. Next, the confidence score of the event is calculated according to the dimensional features. The dimensional features here refer to multiple feature dimensions, which can be the time relationship of the event, the geographical location, the authority of the participants, the influence of the event, etc. By calculating and synthesizing these features, a confidence score can be generated for each event. The confidence score is used to measure the credibility or reliability of the event. Events with high scores are considered to have higher confidence and are therefore more worthy of attention and consideration. Based on the calculation of the confidence score, the events can be further screened, sorted or prioritized. This helps to determine which events are most likely to occur or are most important so that they can be focused on in subsequent analysis, decision-making or action. In summary, by sorting candidate events according to their validity values ​​and calculating the confidence scores of events based on the features of each dimension, we can help find the most valuable and credible events and provide clearer and more reliable information for subsequent processing and decision-making.

[0198] A first candidate event with the highest confidence is selected and output as the first event data.

[0199] In this embodiment, through the preset event recognition model, hot events with news value can be effectively detected quickly from massive social data.

[0200] In some possible implementations of the present invention, the method for constructing the first event recognition model includes:

[0201] Constructing a training data set for first event recognition and annotating event elements;

[0202] Use a sequence labeling model (such as a conditional random field model) to train an event feature extraction model;

[0203] Dictionaries and rule libraries related to the first event are collected, keyword matching and rule matching are performed, and the first event recognition model is obtained by combining the event element extraction model.

[0204] Specifically include:

[0205] Define the representation structure of events, which typically include dimensions such as subject entity, event verb, time and place, and impact results;

[0206] Collect and annotate training data, identify event elements in text, and build a dataset for event extraction;

[0207] Preprocess text data to perform preliminary processing such as word segmentation, part-of-speech tagging, and named entity recognition;

[0208] Use sequence labeling models such as conditional random fields (CRF) to train event feature extraction models with event representation structures as labels;

[0209] Construct feature engineering to extract text features, such as part of speech and dependency syntactic parsing features, to enhance the model's ability to distinguish;

[0210] Collect named entity libraries and event thesaurus in related fields to enrich model features;

[0211] Train multiple feature extraction sub-models, such as subject recognition, event word recognition, etc.

[0212] Design rule-based event filters to remove invalid events that do not meet requirements;

[0213] Use deep learning models for end-to-end event extraction and classification;

[0214] Combine multiple models to build modules such as event candidate generation, event element extraction, event filtering, and event classification to obtain the first event recognition model;

[0215] The event recognition effect of the first event recognition model is evaluated on the test set, and the accuracy of the model is iteratively optimized.

[0216] In this embodiment, by constructing an event recognition model including multiple modules, accurate event extraction and classification can be performed on the text, providing structured event data for subsequent applications.

[0217] In some possible implementations of the present invention, the step of performing cluster analysis on the event portrait data to obtain multiple event portrait data groups includes:

[0218] After preprocessing the event portrait data (such as processing missing values, normalization, etc.), event features are extracted;

[0219] Selecting a suitable clustering algorithm (such as K-Means, DBSCAN, etc.) according to the feature dimensions of the event features;

[0220] For the text features in the event features, a vector representation is obtained using an NLP model, and similarities between texts are calculated based on the vector representation;

[0221] Determining a distance metric (such as Euclidean distance, etc.) for the structured features in the event features;

[0222] According to the similarity and the distance metric, clustering of different granularities is performed to obtain a hierarchical event clustering structure;

[0223] According to the event clustering structure, the center points of different clusters are analyzed to summarize the main characteristic description data of each cluster;

[0224] Calculating the intra-class aggregation degree and inter-class isolation degree of each cluster according to the main feature description data, evaluating the pros and cons of different clustering results, and selecting the first clustering result that meets the preset conditions (such as the optimal clustering result);

[0225] The event portrait data group is obtained according to the first clustering result.

[0226] In this embodiment, through scientific clustering methods, patterns in event data can be discovered and event portrait groups that conform to reality can be generated to provide support for subsequent analysis tasks.

[0227] In some possible implementations of the present invention, the step of generating a corresponding data clustering analysis model for each event portrait data group to obtain multiple event data analysis models includes:

[0228] Analyze the group characteristics of each of the event portrait data groups (such as event type, theme, impact, etc.);

[0229] According to the types of the different groups of features, select appropriate feature engineering methods to perform feature extraction and conversion;

[0230] For text data, use the NLP model to obtain word vector representation to calculate the text semantic similarity;

[0231] For structured data features, build custom feature vectors (you can also use Embedding technology to learn feature representations);

[0232] Fit different clustering algorithms and evaluate the clustering effect of each algorithm on a specific event portrait group;

[0233] Adjust parameters and optimize clustering algorithms to improve clustering accuracy and quality (such as optimizing intra-class aggregation and inter-class separation);

[0234] Build semi-supervised clustering models and use data samples to verify and adjust cluster boundaries;

[0235] Use labeled data to train a custom clustering neural network in a supervised manner;

[0236] Evaluate the explanatory and predictive abilities of different models for groups of event profiles;

[0237] The optimal model is selected as the cluster analysis model corresponding to the event portrait group to obtain the event data analysis model.

[0238] In this embodiment, through this targeted model construction, the data patterns of different types of events can be more accurately characterized, thereby improving the effectiveness of subsequent management decisions.

[0239] In some possible implementations of the present invention, the step of determining the corresponding first event data analysis model from a plurality of event data analysis models according to the attribute data in the first event data includes:

[0240] Extracting event structured attribute features (such as event type, subject, impact, etc.) from the first event data;

[0241] Representing the event structured attribute features as a structured feature vector;

[0242] Calculating the similarity between the feature vector and the central vectors of different event portrait groups;

[0243] According to the similarity sorting, the optimal event portrait data group that best matches the structured attribute characteristics of the event is selected;

[0244] According to the optimal event portrait data group, determining a corresponding basic event data analysis model from a plurality of the event data analysis models;

[0245] The first event data is used to adjust the basic event data analysis model using the transfer learning technology to obtain the first event data analysis model. Specifically, the feature vector is input, and the basic event data analysis model is used to perform predictive analysis; the structure and parameters of the basic event data analysis model are adjusted according to the analysis results.

[0246] In this embodiment, by intelligently matching the model with the event attributes, the preset model that is most suitable for analyzing the event data can be dynamically determined, so that subsequent analysis can obtain more accurate results.

[0247] In some possible implementations of the present invention, the step of analyzing the first event data according to the first event data analysis model to obtain a first analysis result includes:

[0248] Input the first event data, including text description, entity, time, location and other dimensional information;

[0249] In the selected first event data analysis model, text features are extracted and represented as vectors;

[0250] Encode the structured data features and input them into the structured feature analysis module of the model;

[0251] The model fuses the text vector and the structured feature vector to obtain a comprehensive feature expression of the event;

[0252] Use classification / clustering algorithms such as random forest and SVM to predict the impact of events;

[0253] Generate semantic reasoning about the development of events, and use knowledge graphs and reasoning to analyze the entity relationships involved in the events and their impacts;

[0254] Conduct sentiment analysis, intention judgment, etc. to analyze the public’s reaction to the event;

[0255] Comprehensive analysis of each module is performed to generate a comprehensive analysis report of the event as a result;

[0256] Based on the feedback from the first events, the analysis and generation functions of the model are continuously optimized.

[0257] In this embodiment, the designed analysis model can be used to comprehensively analyze event data and generate valuable analysis results.

[0258] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0259] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0260] In the several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above-mentioned units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0261] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0262] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0263] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the above-mentioned methods of each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or CD-ROM and other media that can store program codes.

[0264] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (English: Read-Only Memory, abbreviated as: ROM), a random access memory (English: Random Access Memory, abbreviated as: RAM), a magnetic disk or an optical disk, etc.

[0265] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0266] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions without departing from the spirit and scope of the present invention, and can make various changes and modifications, including the combination of the above-mentioned different functions and implementation steps, including software and hardware implementation methods, all of which are within the scope of protection of the present invention.

Claims

1. A social media data aggregation and analysis system, characterized in that: include: cloud servers and social media platforms; The cloud server is configured as follows: Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships; Extracting historical feature data and historical event data of the social media platform from the historical data; Generate multiple social media platform portrait data based on the historical feature data; Generate multiple event portrait data according to the historical event data; Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data; Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups; Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models; Acquire first social media data from the social media platform according to the first data acquisition model; identifying first event data from the first social media data; Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data in the first event data; Analyze the first event data according to the first event data analysis model to obtain a first analysis result; The step of obtaining historical data of multiple social media platforms and marking the historical data with the unique identifier and time relationship of the social media platform to establish an internal association relationship includes: Obtain developer documentation of the social media platform, and extract the API interface and access permissions provided by the social media platform; Acquire the historical data according to the API interface and access rights of the social media platform; When acquiring the historical data, marking the historical data using the unique identifier of the social media platform and the time relationship; The historical data is stored in a database, and association modeling is performed on the historical data according to the unique identifier and time relationship of the social media platform to obtain the association relationship; The step of generating multiple social media platform portrait data according to the historical feature data includes: Determining the types of features included in the social media platform profile data; According to the feature type, feature engineering is performed on the historical feature data to obtain first feature data; Based on the first characteristic data, a plurality of social media platform portrait data are generated.

2. A social media data aggregation analysis method, characterized in that: include: Acquire historical data of multiple social media platforms, and mark the historical data with unique identifiers and time relationships of the social media platforms to establish internal association relationships; Extracting historical feature data and historical event data of the social media platform from the historical data; Generate multiple social media platform portrait data based on the historical feature data; Generate multiple event portrait data according to the historical event data; Establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data; Performing cluster analysis on the event portrait data to obtain multiple event portrait data groups; Generate a corresponding data clustering analysis model for each of the event portrait data groups to obtain multiple event data analysis models; Acquire first social media data from the social media platform according to the first data acquisition model; identifying first event data from the first social media data; Determine a corresponding first event data analysis model from a plurality of the event data analysis models according to the attribute data of the first event data; Analyze the first event data according to the first event data analysis model to obtain a first analysis result; The step of obtaining historical data of multiple social media platforms and marking the historical data with the unique identifier and time relationship of the social media platform to establish an internal association relationship includes: Obtain developer documentation of the social media platform, and extract the API interface and access permissions provided by the social media platform; Acquire the historical data according to the API interface and access rights of the social media platform; When acquiring the historical data, marking the historical data using the unique identifier of the social media platform and the time relationship; The historical data is stored in a database, and association modeling is performed on the historical data according to the unique identifier and time relationship of the social media platform to obtain the association relationship; The step of generating multiple social media platform portrait data according to the historical feature data includes: Determining the types of features included in the social media platform profile data; According to the feature type, feature engineering is performed on the historical feature data to obtain first feature data; Based on the first characteristic data, a plurality of social media platform portrait data are generated.

3. The social media data aggregation analysis method according to claim 2, characterized in that: The step of generating a plurality of event portrait data according to the historical event data comprises: Use natural language processing technology to extract event elements from the historical event data to obtain event information including event subject, time, location, cause, development process, and transmission path; Applying knowledge graph technology to process the event information and construct an event relationship graph; Use sentiment analysis technology and intention recognition technology to analyze the historical event data and determine the emotional tendency data of participants and the public; Using a social network analysis algorithm to identify the spread scope and spread model of the event from the historical event data; Using multimodal analysis technology to associate text data and multimedia data in the historical event data to build an event knowledge base; A structured event portrait is generated according to the event relationship diagram, the emotional tendency data, the propagation scope, the propagation model, and the event knowledge base to obtain the event portrait data.

4. The social media data aggregation analysis method according to claim 3 is characterized in that: The step of establishing a first data acquisition model for each of the social media platforms according to the social media platform portrait data includes: Determine the user portrait, content portrait, interaction portrait, platform positioning and scenario, security policies and restrictions of the social media platform based on the social media platform portrait data; The first data acquisition model is determined according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policies and restrictions.

5. The social media data aggregation analysis method according to claim 4, characterized in that: The step of determining the first data acquisition model according to the user portrait, the content portrait, the interaction portrait, the platform positioning and scenario, and the security policy and restrictions includes: According to the user portrait, determine the data collection granularity for different user characteristics; Determine the data type / category to be collected based on the content portrait; Determine the collected data of core node users based on the interaction portrait; According to the platform positioning and scenarios, customize the data collection strategy and determine the media type, time range, and collection frequency to be collected; Determine data access rights and authorization mechanisms based on the security policies and restrictions; The data acquisition method is determined based on the technical attributes of the social media platform.

6. The social media data aggregation analysis method according to claim 5, characterized in that: The step of acquiring first social media data from the social media platform according to the first data acquisition model includes: Parsing the collection requirements, user groups, collection keywords, collection methods, and collection frequency of the social media platform from the first data acquisition model; The first social media data is acquired according to the acquisition requirement, the user group, the acquisition keyword, the acquisition method, and the acquisition frequency.

7. The social media data aggregation analysis method according to claim 6, characterized in that: The step of identifying first event data from the first social media data comprises: Inputting the first social media data into a preset first event recognition model; The first event recognition model applies a text similarity algorithm to compare the similarity between the short text extracted from the first social media data and the historical event brief introduction to obtain a candidate event set; Using named entity recognition and entity linking technology, identifying event subjects and relationships between event subjects from the first social media data; By means of the knowledge graph, the relationship between event subjects is analyzed to determine the validity value of each candidate event in the candidate event set; Clustering each of the candidate events, and filtering noise data using statistical characteristics of the clusters; Determine the time range and geographical scope of the event by combining the time and location information identified from the first social media data; Sorting each of the candidate events according to the validity value, and calculating the confidence score of the event according to each dimensional feature; A first candidate event with the highest confidence is selected and output as the first event data.

8. The social media data aggregation analysis method according to claim 7, characterized in that: The method for constructing the first event recognition model includes: Constructing a training data set for first event recognition and annotating event elements; Use the sequence labeling model to train the event feature extraction model; Dictionaries and rule libraries related to the first event are collected, keyword matching and rule matching are performed, and the first event recognition model is obtained by combining the event element extraction model.

Images