Social network abnormal information propagation tracking system based on potential propagation set

Abstract

The invention discloses a social network abnormal information propagation tracking system based on a potential propagation set, and relates to the technical field of social network security. In order to solve the problems in the prior art that subjectivity and errors possibly exist, extraction and adjustment of propagation parameters are affected, so that prediction accuracy is affected, the propagation condition of network hotspot messages needs to be predicted through manual intervention, and the capability of the network hotspot messages in the aspects of automatic prediction and monitoring is limited. The social network abnormal information propagation tracking system based on the potential propagation set comprises a data acquisition unit, a potential propagation set generation unit, a propagation path tracking unit and an early warning unit, by analyzing the interaction relationship and the information propagation rule of the nodes in the social network, the potential propagation set is identified, and the propagation path of the information is predicted and tracked by using the potential propagation set, so that the propagation of the abnormal information is discovered and dealt with in time, enterprises and organizations are helped to discover and process the abnormal information in time, and the health and safety of the network space are maintained.

Description

Technical Field

[0001] The present invention relates to the technical field of social network security, and in particular to a social network abnormal information propagation tracking system based on potential propagation sets. Background Art

[0002] With the rapid development of social networks, the speed and scope of information dissemination have greatly increased, but it has also brought about the problem of abnormal information dissemination. Regarding the tracking of abnormal information dissemination in social networks based on potential dissemination sets, a Chinese patent with publication number CN106844697B discloses a method for tracking information dissemination in social networks based on potential dissemination sets, including: establishing a SIR model based on potential dissemination sets; obtaining network messages from social networks, clustering and grouping them, and establishing a classification database of network hot messages; selecting network hot messages from the classification database, sampling the social network, and obtaining dissemination parameters; obtaining new network hot messages, comparing them with the classification database of network hot messages, selecting the closest network hot message type, and extracting the obtained dissemination parameters as reference dissemination parameters of new network hot messages; adjusting the reference dissemination parameters, further predicting the dissemination of the network hot messages and performing manual intervention.

[0003] Although the above patent avoids the problem of repeated counting, it still has the following problems:

[0004] In the prior art, after obtaining a new network hot spot message, it is necessary to compare it with the existing network hot spot messages and select the closest type. There may be subjectivity and errors, which will affect the extraction and adjustment of propagation parameters, thereby affecting the accuracy of the prediction. In addition, manual intervention is required to predict the propagation of network hot spot messages, which limits its ability in automatic prediction and monitoring. Summary of the invention

[0005] The purpose of the present invention is to provide a social network abnormal information propagation tracking system based on potential propagation sets, which identifies potential propagation sets by analyzing the interactive relationships and information propagation rules of nodes in social networks, and uses potential propagation sets to predict and track the propagation paths of information, so as to timely discover and respond to the propagation of abnormal information, help enterprises and organizations to timely discover and handle abnormal information, maintain the health and security of cyberspace, and solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The social network abnormal information propagation tracking system based on potential propagation sets includes:

[0008] A data collection unit, used to crawl required target data from a social network base, wherein the target data includes user interaction information, published content, and forwarding relationship data;

[0009] A potential propagation set generation unit is used to analyze the crawled target data using the constructed propagation model, identify the potential propagation sets in the target data, and screen out the potential propagation sets with abnormal characteristics;

[0010] A propagation path tracking unit is used to track the potential propagation set with abnormal characteristics in real time based on the identified potential propagation set, record its propagation path, analyze the key role of key nodes in the potential propagation set in information propagation, and predict the diffusion trend of information;

[0011] The early warning unit is used to compare the analysis and prediction results of the propagation path tracking unit with the key nodes of the normal information propagation mode, detect the propagation of abnormal information based on the comparison results, and issue an early warning based on the propagation path information of the abnormal information, and notify the staff to take countermeasures.

[0012] Furthermore, the data acquisition unit comprises:

[0013] Obtaining user behavior and relationship data from each social network, and obtaining data types of the user behavior and relationship data;

[0014] Preprocessing the user behavior and relationship data corresponding to the data type, eliminating redundant data, and obtaining processed user behavior and relationship data;

[0015] The user behavior and relationship data are filtered according to the effective data transmission status of the social network to obtain target data, wherein the target data corresponds to the effective data transmission status one by one.

[0016] Further, preprocessing is performed on the user behavior and relationship data corresponding to the data type to remove redundant data and obtain processed user behavior and relationship data, including:

[0017] Inputting the user behavior and relationship data corresponding to the data type into a data preprocessing unit;

[0018] After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are duplicate data records in the data;

[0019] When there are duplicate data records in the user behavior and relationship data corresponding to the data type, extracting the generation time corresponding to the duplicate data record;

[0020] After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are abnormal values ​​in the data;

[0021] When an abnormal value exists in the user behavior and relationship data corresponding to the data type, extracting a generation time corresponding to the abnormal value;

[0022] Using the generation time corresponding to the repeated data record, obtaining the middle time in the distribution of the generation time corresponding to the repeated data record;

[0023] Compare the generation time corresponding to the abnormal value with the middle time corresponding to the repeated data record to determine whether the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record satisfies a preset constraint condition, wherein the constraint condition is as follows:

[0024]

[0025] Among them, T c represents the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record; λ represents a constant coefficient; T y represents the generation time corresponding to the abnormal value; T h Indicates the generation time of the last data record of the repeated data record; T x Indicates the generation time of the first data record of the repeated data record; n indicates the number of repeated data records; T gi Indicates the generation time corresponding to the i-th repeated data record contained in the repeated data records;

[0026] When the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record meets the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data;

[0027] When the time difference between the generation moment corresponding to the abnormal value and the middle moment corresponding to the repeated data record does not satisfy the preset constraint condition, the repeated data record and the abnormal value are both removed.

[0028] Further, when the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record satisfies the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data, including:

[0029] When the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the data difference value between the abnormal value and the repeated data record is retrieved;

[0030] The reference time corresponding to the replacement value of the abnormal value is obtained by using the data difference value between the abnormal value and the repeated data record; wherein the replacement value is obtained by the following formula:

[0031]

[0032] Where X represents the reference time corresponding to the replacement value; x represents the value coefficient, and when T y -T x <T c When x=1, when T y -T x >T c When x = -0.87; when T y -T x =T c When x=0;

[0033] Filtering the data parameters corresponding to the time closest to the reference time in the user behavior and relationship data, and using the data parameters corresponding to the time closest to the reference time as replacement values;

[0034] replacing the abnormal value with the replacement value;

[0035] After the abnormal value is completely replaced by the replacement value, the duplicate data record is retrieved and the duplicate data record is deleted.

[0036] Furthermore, the potential propagation set generating unit includes:

[0037] A data processing module is used to process target data in a large-scale social network, build a classification database, and store the acquired target data in the classification database based on data type classification;

[0038] The potential propagation set identification module is used to integrate users with similar behavioral characteristics based on the processing results of the data processing module, identify potential propagation sets, match the extracted features with known abnormal information propagation patterns, and identify potential propagation sets with anomalies.

[0039] Furthermore, the data processing module includes:

[0040] A parameter optimization and adjustment submodule, used to obtain historical data of the social network, determine the propagation parameters of the target data based on the propagation parameters of the historical data, and adjust the parameters of the real-time changing data of the social network according to the matching results;

[0041] The scalability optimization submodule is used to build a distributed computing framework, process and analyze target data based on the distributed computing framework, and establish an index query optimization function based on the analysis results.

[0042] Furthermore, the potential propagation set identification module is specifically:

[0043] Extracting behavior features from the user's behavior data based on the data type of the target data, and grouping the user data corresponding to the behavior features based on a clustering algorithm;

[0044] Extracting connection relationships between users from a social network, and determining association relationships between users in each group based on the connection relationships;

[0045] Analyze the growth dynamics of each user group based on time series to identify whether there is an abnormally growing user group in the user group;

[0046] A known abnormal information propagation pattern library is established, each identified potential propagation set is matched with the abnormal information propagation pattern library, and abnormal pattern recognition is performed on each identified potential propagation set.

[0047] Furthermore, the propagation path tracking unit comprises:

[0048] The real-time monitoring and prediction module is used to track and predict the propagation of the potential propagation set with abnormalities identified by the potential propagation set identification module in real time based on the stream data processing technology, analyze the propagation trend based on the time series, and predict the propagation situation in the future;

[0049] A social network structure analysis module is used to analyze the community structure in the social network based on the connection relationship between users and determine the propagation nodes of the potential propagation set in the community structure;

[0050] The message adaptability analysis module is used to extract key features from the message content corresponding to the potential propagation set, determine the message type and propagation characteristics of the message content, and automatically identify and classify different types of message content based on the key features.

[0051] Furthermore, the message adaptability analysis module is specifically:

[0052] Create a filtering word library, and perform word segmentation on the body and title of the message content to obtain a number of extracted words, determine the word features of each extracted word and the sentence features of each extracted word in the body and title;

[0053] Based on the constructed synonym dictionary and the word features of each extracted word and the sentence features of each extracted word in the text and the title, determine the similarity between the extracted words, and perform word screening based on the similarity;

[0054] The words are cleaned according to the statistical features of parts of speech to obtain the main body of the message content, the message type of the message content is determined based on the main body of the message content, the key features in each message content are determined using the writing weight, and the propagation characteristics of the message content are determined based on the key features.

[0055] Furthermore, the early warning unit includes:

[0056] The abnormal warning module is used to automatically trigger the warning mechanism according to the prediction results of the real-time monitoring and prediction module, determine the warning level according to the severity and impact scope of the abnormal information, and transmit the warning information to the management terminal based on the Internet of Things;

[0057] The visualization display module is used to provide a human-computer interaction interface, display the real-time status of the social network and the propagation path of abnormal information in real time, and generate reports based on the visualization results.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] 1. By filtering the effective data transmission status and obtaining the target data, the real-time and effectiveness of the data can be ensured, and the timeliness of tracking the spread of abnormal information can be improved. By monitoring the effective data transmission status of social networks and filtering the effective data transmission status to obtain the target data, it can better reflect the actual situation of user behavior and information dissemination and improve the utility value of the data.

[0060] 2. By analyzing the interactive relationships and information dissemination patterns of nodes in social networks, potential dissemination sets are identified, and potential dissemination sets are used to predict and track the dissemination paths of information, so as to timely discover and respond to the dissemination of abnormal information, help enterprises and organizations to timely discover and deal with abnormal information, maintain the health and security of cyberspace, reduce the burden of manual monitoring through automated early warning mechanisms, improve the efficiency of monitoring the dissemination of abnormal information, take appropriate intervention measures according to the early warning level, effectively prevent the further dissemination of abnormal information, and reduce potential harm and impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a module diagram of the social network abnormal information propagation tracking system based on potential propagation sets of the present invention;

[0062] Figure 2 It is a flow chart of the algorithm of the potential propagation set identification module of the present invention;

[0063] Figure 3This is a workflow diagram of the potential propagation set identification module of the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] In order to solve the technical problems in the existing technology that there may be subjectivity and errors, which affect the extraction and adjustment of propagation parameters, thereby affecting the accuracy of prediction, and requiring manual intervention to predict the propagation of network hotspot messages, limiting its ability in automatic prediction and monitoring, please refer to Figure 1-3 , this embodiment provides the following technical solutions:

[0066] The social network abnormal information propagation tracking system based on potential propagation sets includes:

[0067] A data collection unit is used to crawl required target data from the social network base, wherein the target data includes user interaction information, published content and forwarding relationship data. Such data can be obtained by crawlers and other technical means, including various media content such as text, pictures, and videos;

[0068] In this embodiment, the data acquisition unit includes:

[0069] Obtain user behavior and relationship data from each social network, and at the same time, obtain the data type of the user behavior and relationship data, including user posting, commenting, forwarding and other behavior data, as well as relationship data between users;

[0070] Preprocessing the user behavior and relationship data corresponding to the data type, eliminating redundant data, obtaining processed user behavior and relationship data, reducing data noise, improving data quality, and providing a more accurate data basis for subsequent abnormal information detection;

[0071] The user behavior and relationship data are filtered according to the effective data transmission status of the social network to obtain the target data, wherein the target data corresponds to the effective data transmission status one by one. By filtering the effective data transmission status and obtaining the target data, the real-time and effectiveness of the data can be ensured, and the timeliness of abnormal information propagation tracking can be improved;

[0072] In this embodiment, by monitoring the effective data transmission status of the social network and filtering the effective data transmission status to obtain the target data, it is possible to better reflect the real situation of user behavior and information dissemination and improve the utility value of the data;

[0073] A potential propagation set generation unit is used to analyze the crawled target data using the constructed propagation model, identify the potential propagation sets in the target data, and screen out the potential propagation sets with abnormal characteristics. The propagation model is selected or optimized according to actual needs to accurately identify the potential propagation sets to the greatest extent possible.

[0074] A propagation path tracking unit is used to track the potential propagation set with abnormal characteristics in real time based on the identified potential propagation set, record its propagation path, analyze the key role of key nodes in the potential propagation set in information propagation, and predict the diffusion trend of information. The key nodes may be key opinion leaders, topic spreaders or other influential users. Specifically, the tracking method may adopt a graph-based tracking algorithm, build a social network graph based on user behavior data, and use the graph algorithm to track the path of the potential propagation set;

[0075] The early warning unit is used to compare the analysis and prediction results of the propagation path tracking unit with the key nodes of the normal information propagation pattern, and detect the propagation of abnormal information based on the comparison results. For example, the sudden appearance of a large number of forwarding, commenting or sharing behaviors may indicate the emergence of abnormal information, and issue an early warning based on the propagation path information of the abnormal information, notifying the staff to take countermeasures, such as restricting the spread of information, deleting malicious content or taking other appropriate measures.

[0076] In this embodiment, the early warning unit includes:

[0077] The abnormal warning module is used to automatically trigger the warning mechanism according to the prediction results of the real-time monitoring and prediction module, determine the warning level according to the severity and impact range of the abnormal information, and transmit the warning information to the management terminal based on the Internet of Things to intervene in the abnormal propagation in time;

[0078] The visualization module is used to provide a human-computer interaction interface, display the real-time status of social networks and the propagation path of abnormal information in real time, and generate reports based on the visualization results for further analysis or decision-making, and display the propagation status and trends of hot network messages in real time;

[0079] In this embodiment, by analyzing the interactive relationships and information propagation rules of nodes in a social network, potential propagation sets are identified, and potential propagation sets are used to predict and track the propagation path of information, so as to timely discover and respond to the spread of abnormal information, help enterprises and organizations to timely discover and handle abnormal information, maintain the health and security of cyberspace, reduce the burden of manual monitoring through an automated early warning mechanism, improve the efficiency of monitoring the spread of abnormal information, take corresponding intervention measures according to the early warning level, effectively prevent the further spread of abnormal information, and reduce potential harm and impact.

[0080] Specifically, preprocessing the user behavior and relationship data corresponding to the data type, removing redundant data, and obtaining processed user behavior and relationship data includes:

[0081] Inputting the user behavior and relationship data corresponding to the data type into a data preprocessing unit;

[0082] After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are duplicate data records in the data;

[0083] When there are duplicate data records in the user behavior and relationship data corresponding to the data type, extracting the generation time corresponding to the duplicate data record;

[0084] After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are abnormal values ​​in the data;

[0085] When an abnormal value exists in the user behavior and relationship data corresponding to the data type, extracting a generation time corresponding to the abnormal value;

[0086] Using the generation time corresponding to the repeated data record, obtaining the middle time in the distribution of the generation time corresponding to the repeated data record;

[0087] Compare the generation time corresponding to the abnormal value with the middle time corresponding to the repeated data record to determine whether the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record satisfies a preset constraint condition, wherein the constraint condition is as follows:

[0088]

[0089] Among them, T c represents the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record; λ represents a constant coefficient; T y represents the generation time corresponding to the abnormal value; T hIndicates the generation time of the last data record of the repeated data record; T x Indicates the generation time of the first data record of the repeated data record; n indicates the number of repeated data records; T gi Indicates the generation time corresponding to the i-th repeated data record contained in the repeated data records;

[0090] When the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record meets the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data;

[0091] When the time difference between the generation moment corresponding to the abnormal value and the middle moment corresponding to the repeated data record does not satisfy the preset constraint condition, the repeated data record and the abnormal value are both removed.

[0092] The technical effect of the above technical solution is: the above technical solution cleans the user behavior and relationship data through the data preprocessing unit, effectively removes duplicate data records, thereby reducing data redundancy and improving data quality and accuracy. This is crucial for subsequent data analysis and mining.

[0093] The above technical solution not only focuses on the processing of duplicate data, but also takes into account the processing of outliers. By checking the outliers in the data and comparing the generation time of the outliers with the middle time of the duplicate data records, it is determined whether these outliers need to be processed. This method helps to maintain the integrity and consistency of the data while avoiding the adverse effects of outliers on subsequent analysis. At the same time, the constraint design in the above technical solution has certain flexibility and scalability. By introducing the constant coefficient λ, the constraints can be adjusted according to actual needs to adapt to different application scenarios and data characteristics. This makes the above technical solution more versatile and adaptable when processing different types of data.

[0094] On the other hand, through the automated data processing process, the above technical solution can efficiently process a large amount of user behavior and relationship data, reduce manual intervention and errors, and improve the speed and accuracy of data processing. This is of great significance for real-time or near real-time data analysis and decision support. By effectively processing duplicate data and outliers, the above technical solution enhances the reliability of data and provides more reliable basic data for subsequent data analysis and mining. This helps to improve the accuracy and effectiveness of analysis and provide strong support for the decision-making of enterprises or organizations.

[0095] In summary, the above technical solutions have significant technical effects in data cleaning, redundancy removal, outlier processing, etc., which help to improve the quality and accuracy of data, enhance data reliability, and provide strong guarantees for data analysis and decision support.

[0096] Specifically, when the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data, including:

[0097] When the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the data difference value between the abnormal value and the repeated data record is retrieved;

[0098] The reference time corresponding to the replacement value of the abnormal value is obtained by using the data difference value between the abnormal value and the repeated data record; wherein the replacement value is obtained by the following formula:

[0099]

[0100] Where X represents the reference time corresponding to the replacement value; x represents the value coefficient, and when T y -T x <T c When x=1, when T y -T x >T c When x = -0.87; when T y -T x =T c When x=0;

[0101] Filtering the data parameters corresponding to the time closest to the reference time in the user behavior and relationship data, and using the data parameters corresponding to the time closest to the reference time as replacement values;

[0102] replacing the abnormal value with the replacement value;

[0103] After the abnormal value is completely replaced by the replacement value, the duplicate data record is retrieved and the duplicate data record is deleted.

[0104] The technical effect of the above technical solution is: when the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the solution will further process these abnormal values. By calculating the data difference value between the abnormal value and the repeated data record, the relationship between the abnormal value and other data can be more accurately understood, providing a basis for subsequent replacement operations.

[0105] At the same time, using the data difference value between the outlier and the duplicate data record, the scheme obtains a reference time for the replacement value. This replacement strategy based on data difference can ensure that the replaced value is closer to the actual situation and reduce the interference of the replacement operation on the data set. On the other hand, the value coefficient x in the above technical scheme is dynamically adjusted according to the relationship between the time difference and the preset threshold, which increases the flexibility and adaptability of the scheme. By screening the data parameters corresponding to the moment closest to the reference moment in the user behavior and relationship data as the replacement value, the scheme can ensure the accuracy and rationality of the replacement operation. Doing so can reduce the deviation between the replacement value and the original data set and improve the overall quality of the data. After completing the replacement of the outlier, the scheme will further eliminate duplicate data records. This helps to reduce redundant information in the data set and improve the clarity and analyzability of the data.

[0106] In summary, this technical solution improves the quality and accuracy of user behavior and relationship data through an outlier processing strategy based on data differences and the elimination of duplicate data. This processing method can reduce noise and interference in the data set and provide more reliable basic data for subsequent data analysis and mining.

[0107] In this embodiment, the potential propagation set generating unit includes:

[0108] The data processing module is used to process the target data in the large-scale social network, cluster them quickly and effectively, reduce the complexity of the data, build a classification database, and store the acquired target data in the classification database based on the data type classification. With the addition of new messages, the classification database can be dynamically updated to avoid large-scale data reorganization;

[0109] In this embodiment, the data processing module includes:

[0110] A parameter optimization and adjustment submodule is used to obtain historical data of the social network, determine the propagation parameters of the target data based on the propagation parameters of the historical data, reduce the need for manual intervention, and adjust the parameters of the real-time changing data of the social network according to the matching results to improve the prediction accuracy;

[0111] The scalability optimization submodule is used to build a distributed computing framework, process and analyze the target data based on the distributed computing framework, and establish an index query optimization function based on the analysis results to improve data processing speed and query efficiency;

[0112] The potential propagation set identification module is used to integrate users with similar behavioral characteristics based on the processing results of the data processing module, identify potential propagation sets, find user groups with close relationships in social networks, and these groups may become propagation sets of abnormal information. The extracted features are matched with known abnormal information propagation patterns to identify potential propagation sets with anomalies.

[0113] In this embodiment, the potential propagation set identification module is specifically:

[0114] Extracting behavioral features from the user's behavioral data based on the data type of the target data, the behavioral features include user posting frequency, comment activity, forwarding behavior, etc., and grouping the user data corresponding to the behavioral features based on a clustering algorithm to understand the user's behavior pattern and participation in the social network;

[0115] Extract the connection relationships between users from the social network. The connection relationships include following relationships, friend relationships, interactive behaviors, etc. Based on the connection relationships, determine the association relationships between users in each group, which can reflect the closeness between users and the information dissemination path, and deeply understand the internal structure and interaction mode of the user group;

[0116] Based on the time series, we analyze the growth dynamics of each user group to identify whether there are any abnormally growing user groups. By monitoring the size changes of the user groups, we can identify whether there are any abnormally growing user groups. Abnormal growth may indicate that the group is more active and influential in the dissemination of abnormal information.

[0117] Establish a library of known abnormal information propagation patterns, match each identified potential propagation set with the abnormal information propagation pattern library, and perform abnormal pattern recognition on each identified potential propagation set. By comparing similarities and pattern matching, the module can perform abnormal pattern recognition on each identified potential propagation set, which helps to determine whether the group is involved in the spread of abnormal information and determine its potential degree of harm.

[0118] In this embodiment, based on methods such as clustering algorithms and time series analysis, the potential propagation set identification module can more accurately identify potential propagation sets, reduce the possibility of misjudgment and missed judgment, monitor the dynamic changes of user behavior and relationship data in real time, promptly discover abnormally growing user groups, and take corresponding intervention measures to effectively control the further spread of abnormal information. Algorithm optimization and parameter adjustment can be performed according to actual needs and data scale to improve the performance and accuracy of the module. According to the characteristics and needs of different social networks, the module can provide customized functions and settings to meet the needs of abnormal information propagation tracking in different scenarios, reduce the cost and frequency of manual intervention, and improve the efficiency of abnormal information propagation tracking.

[0119] In this embodiment, the propagation path tracking unit includes:

[0120] The real-time monitoring and prediction module is used to track and predict the propagation of the potential propagation set with abnormalities identified by the potential propagation set identification module in real time based on the stream data processing technology, analyze the propagation trend based on the time series, and predict the propagation situation in the future;

[0121] The social network structure analysis module is used to analyze the community structure in the social network based on the connection relationship between users, determine the propagation nodes of the potential propagation set in the community structure, understand how information propagates among different communities, and consider the relationship between nodes, such as attention, forwarding, and likes, to more accurately predict the information propagation path;

[0122] The message adaptability analysis module is used to extract key features, such as sentiment and theme, from the message content corresponding to the potential propagation set, determine the message type and propagation characteristics of the message content, and automatically identify and classify different types of message content using classification algorithms such as decision trees and support vector machines based on the key features;

[0123] In this embodiment, the message adaptability analysis module is specifically:

[0124] Create a filtering word library, and perform word segmentation on the body and title of the message content to obtain a number of extracted words, determine the word features of each extracted word and the sentence features of each extracted word in the body and title;

[0125] Based on the constructed synonym dictionary and the word features of each extracted word and the sentence features of each extracted word in the text and the title, determine the similarity between the extracted words, and perform word screening based on the similarity;

[0126] The words are cleaned according to the statistical features of parts of speech to obtain the main body of the message content, and the message type of the message content is determined based on the main body of the message content. The writing weight is used to determine the key features in each message content. The key features may include keywords, emotional tendencies, topic focus, etc., which can reflect the core information and dissemination intention of the message. The dissemination characteristics of the message content are determined based on the key features.

[0127] In this embodiment, the message content is segmented, feature extracted and similarity analyzed. The module can more accurately identify and extract key features, reduce the possibility of misjudgment and missed judgment, and promptly discover the propagation trend of abnormal information and changes in key features, providing support for real-time monitoring and intervention. Based on the main body of the message content, the module determines the message type of the message content, thereby determining the different propagation characteristics and influence of different types of messages, and evaluating indicators such as the message's propagation range, speed, influence and credibility to determine whether the message has the characteristics of abnormal propagation and potential hazards.

[0128] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A social network abnormal information propagation tracking system based on potential propagation sets, characterized by: include: A data collection unit, used to crawl required target data from a social network base, wherein the target data includes user interaction information, published content, and forwarding relationship data; A potential propagation set generation unit is used to analyze the crawled target data using the constructed propagation model, identify the potential propagation sets in the target data, and screen out the potential propagation sets with abnormal characteristics; A propagation path tracking unit is used to track the potential propagation set with abnormal characteristics in real time based on the identified potential propagation set, record its propagation path, analyze the key role of key nodes in the potential propagation set in information propagation, and predict the diffusion trend of information; The early warning unit is used to compare the analysis and prediction results of the propagation path tracking unit with the key nodes of the normal information propagation mode, detect the propagation of abnormal information based on the comparison results, and issue an early warning based on the propagation path information of the abnormal information, and notify the staff to take countermeasures.

2. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 1, characterized in that: Data acquisition unit, including: Obtaining user behavior and relationship data from each social network, and obtaining data types of the user behavior and relationship data; Preprocessing the user behavior and relationship data corresponding to the data type, eliminating redundant data, and obtaining processed user behavior and relationship data; The user behavior and relationship data are filtered according to the effective data transmission status of the social network to obtain target data, wherein the target data corresponds to the effective data transmission status one by one.

3. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 2, characterized in that: Preprocessing the user behavior and relationship data corresponding to the data type, removing redundant data, and obtaining processed user behavior and relationship data includes: Inputting the user behavior and relationship data corresponding to the data type into a data preprocessing unit; After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are duplicate data records in the data; When there are duplicate data records in the user behavior and relationship data corresponding to the data type, extracting the generation time corresponding to the duplicate data record; After receiving the user behavior and relationship data corresponding to the data type, the data preprocessing unit checks whether there are abnormal values ​​in the data; When an abnormal value exists in the user behavior and relationship data corresponding to the data type, extracting a generation time corresponding to the abnormal value; Using the generation time corresponding to the repeated data record, obtaining the middle time in the distribution of the generation time corresponding to the repeated data record; Compare the generation time corresponding to the abnormal value with the middle time corresponding to the repeated data record to determine whether the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record satisfies a preset constraint condition, wherein the constraint condition is as follows: Among them, T c represents the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record; λ represents a constant coefficient; T y represents the generation time corresponding to the abnormal value; T h Indicates the generation time of the last data record of the repeated data record; T x Indicates the generation time of the first data record of the repeated data record; n indicates the number of repeated data records; T gi Indicates the generation time corresponding to the i-th repeated data record contained in the repeated data records; When the time difference between the generation time corresponding to the abnormal value and the middle time corresponding to the repeated data record meets the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data; When the time difference between the generation moment corresponding to the abnormal value and the middle moment corresponding to the repeated data record does not satisfy the preset constraint condition, the repeated data record and the abnormal value are both removed.

4. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 3, characterized in that: When the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the abnormal value and the repeated data record are processed to obtain pre-processed user behavior and relationship data, including: When the time difference between the generation time corresponding to the abnormal value and the generation time corresponding to the repeated data record meets the preset constraint condition, the data difference value between the abnormal value and the repeated data record is retrieved; The reference time corresponding to the replacement value of the abnormal value is obtained by using the data difference value between the abnormal value and the repeated data record; wherein the replacement value is obtained by the following formula: Where X represents the reference time corresponding to the replacement value; x represents the value coefficient, and when T y -T x <T c When x=1, when T y -T x >T c When x = -0.87; when T y -T x =T c When x=0; Filtering the data parameters corresponding to the time closest to the reference time in the user behavior and relationship data, and using the data parameters corresponding to the time closest to the reference time as replacement values; replacing the abnormal value with the replacement value; After the abnormal value is completely replaced by the replacement value, the duplicate data record is retrieved and the duplicate data record is deleted.

5. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 2, characterized in that: Potential propagation set generation unit, including: A data processing module is used to process target data in a large-scale social network, build a classification database, and store the acquired target data in the classification database based on data type classification; The potential propagation set identification module is used to integrate users with similar behavioral characteristics based on the processing results of the data processing module, identify potential propagation sets, match the extracted features with known abnormal information propagation patterns, and identify potential propagation sets with anomalies.

6. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 5, characterized in that: Data processing module, including: A parameter optimization and adjustment submodule, used to obtain historical data of the social network, determine the propagation parameters of the target data based on the propagation parameters of the historical data, and adjust the parameters of the real-time changing data of the social network according to the matching results; The scalability optimization submodule is used to build a distributed computing framework, process and analyze target data based on the distributed computing framework, and establish an index query optimization function based on the analysis results.

7. The social network abnormal information propagation tracking system based on potential propagation sets according to claim 6, characterized in that: Potential propagation set identification module, specifically: Extracting behavior features from the user's behavior data based on the data type of the target data, and grouping the user data corresponding to the behavior features based on a clustering algorithm; Extracting connection relationships between users from a social network, and determining association relationships between users in each group based on the connection relationships; Analyze the growth dynamics of each user group based on time series to identify whether there is an abnormally growing user group in the user group; A known abnormal information propagation pattern library is established, each identified potential propagation set is matched with the abnormal information propagation pattern library, and abnormal pattern recognition is performed on each identified potential propagation set.

8. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 7, characterized in that: Propagation path tracking unit, including: The real-time monitoring and prediction module is used to track and predict the propagation of the potential propagation set with abnormalities identified by the potential propagation set identification module in real time based on the stream data processing technology, analyze the propagation trend based on the time series, and predict the propagation situation in the future; A social network structure analysis module is used to analyze the community structure in the social network based on the connection relationship between users and determine the propagation nodes of the potential propagation set in the community structure; The message adaptability analysis module is used to extract key features from the message content corresponding to the potential propagation set, determine the message type and propagation characteristics of the message content, and automatically identify and classify different types of message content based on the key features.

9. The social network abnormal information propagation tracking system based on potential propagation sets as claimed in claim 8, characterized in that: Message adaptability analysis module, specifically: Create a filtering word library, and perform word segmentation on the body and title of the message content to obtain a number of extracted words, determine the word features of each extracted word and the sentence features of each extracted word in the body and title; Based on the constructed synonym dictionary and the word features of each extracted word and the sentence features of each extracted word in the text and the title, determine the similarity between the extracted words, and perform word screening based on the similarity; The words are cleaned according to the statistical features of parts of speech to obtain the main body of the message content, the message type of the message content is determined based on the main body of the message content, the key features in each message content are determined using the writing weight, and the propagation characteristics of the message content are determined based on the key features.

10. The social network abnormal information propagation tracking system based on potential propagation sets according to claim 9, characterized in that: Early warning unit, including: The abnormal warning module is used to automatically trigger the warning mechanism according to the prediction results of the real-time monitoring and prediction module, determine the warning level according to the severity and impact scope of the abnormal information, and transmit the warning information to the management terminal based on the Internet of Things; The visualization display module is used to provide a human-computer interaction interface, display the real-time status of the social network and the propagation path of abnormal information in real time, and generate reports based on the visualization results.

Images