A public opinion risk evolution prediction system based on AI technology

Abstract

The present invention provides a public opinion risk evolution prediction system based on AI technology, which relates to the field of electronic digital data processing, including a target setting acquisition module, a text analysis and processing module, an event graph analysis module and a risk model prediction module. The target setting acquisition module is used to input a target event and collect network public opinion information related to the target event, the text analysis and processing module is used to parse and obtain risk points in the public opinion information, the event graph analysis module is used to construct a development graph of risk points, and the risk model prediction module uses the development graph information as input to perform risk prediction on the target event. The system can perform digital analysis on the evolution process of public opinion information, predict the probability of occurrence of public opinion accidents, and prepare countermeasures for public opinion in advance based on the predicted probability.

Description

Technical Field

[0001] The present invention relates to the field of electronic digital data processing, and in particular to a public opinion risk evolution prediction system based on AI technology. Background Art

[0002] The development of public opinion is irreversible. As public opinion develops, there is a risk of public opinion accidents, which leads to a lower evaluation of relevant departments. However, the development of public opinion is also sudden and unpredictable. Therefore, a system is needed that can correctly predict the evolution of public opinion so that timely response measures can be taken before the risk breaks out, reducing the possibility of public opinion accidents or the harmfulness of public opinion accidents.

[0003] The foregoing discussion of the background art is intended only to facilitate an understanding of the present invention. This discussion does not acknowledge or admit that any of the material referred to is part of the common general knowledge.

[0004] Now many public opinion prediction systems have been developed. After a lot of searching and reference, it is found that the existing public opinion prediction systems are such as the system disclosed by the publication number CN116383399A. These system methods generally include: collecting the original event information sequence and event-related data, and obtaining the event information sequence after preprocessing; constructing the event-related knowledge graph to obtain the event node representation vector; obtaining the object external industry knowledge representation vector from the event node representation vector, and obtaining the event information representation vector from the event information sequence and the object external industry knowledge representation vector; constructing a risk information transmission network, inputting the risk information transmission network into the time series graph attention network, and obtaining the risk representation vector of the target object at any time; inputting the risk representation vector of the target object at any time into the preset neural network for training, and obtaining the public opinion risk prediction value of the target object at any time. However, this system is based on the state of a certain moment for prediction, lacking the dimensional information of the event development, resulting in the accuracy of the prediction results to be improved. Summary of the invention

[0005] The purpose of the present invention is to propose a public opinion risk evolution prediction system based on AI technology to address the existing deficiencies.

[0006] The present invention adopts the following technical solution:

[0007] A public opinion risk evolution prediction system based on AI technology, including a target setting acquisition module, a text analysis and processing module, an event graph analysis module and a risk model prediction module;

[0008] The target setting collection module is used to input the target event and collect network public opinion information related to the target event, the text analysis processing module is used to analyze and obtain the risk points in the public opinion information, the event map analysis module is used to construct a development map of the risk points, and the risk model prediction module uses the development map information as input to perform risk prediction on the target event;

[0009] The text analysis processing module includes a keyword extraction unit, a risk point matching unit and a risk quantification unit. The keyword extraction unit is used to extract keyword information in the report content, the risk point matching unit is used to match the keyword with the historical case information, and the risk quantification unit is used to calculate the risk value of the report content;

[0010] The event graph analysis module includes an event normalization unit, a risk feature unit and a time management unit. The event normalization unit is used to unify the report content describing the same event into a graph node. The risk feature unit is used to analyze and obtain the risk feature of each graph node. The time management unit is used to identify and manage the time relationship between graph nodes.

[0011] The risk model prediction module includes a graph input unit, a step processing unit and a comprehensive prediction unit. The graph input unit is used to input graph information, the step processing unit is used to analyze and process the development trend between two adjacent graph nodes, and the comprehensive prediction unit is used to generate probability information of the final public opinion accident.

[0012] Further, the risk quantification unit includes a weight calculation processor and a risk calculation processor, the weight calculation processor is used to calculate the risk weight coefficient of each keyword in the report content, and the risk calculation processor is used to calculate the risk value of each report content;

[0013] The weight calculation processor calculates the risk weight coefficient k of the keyword according to the following formula:

[0014] ;

[0015] in, is the number of matching cases containing the corresponding keyword, and m is the number of matching cases;

[0016] Furthermore, the risk calculation processor calculates the risk value Vr of each report content according to the following formula:

[0017] ;

[0018] Among them, N is the number of keywords contained in the report content, k i represents the risk weight coefficient of the ith keyword in the report content;

[0019] Further, the ladder processing unit includes a vector management processor and a ladder calculation processor, the vector management processor is used to receive a node vector and send two adjacent node vectors to the ladder calculation processor, and the ladder calculation processor is used to calculate the ladder change value of the two node vectors;

[0020] The step calculation processor calculates the step change value according to the following formula: :

[0021] ;

[0022] in, is the average risk value in the next node vector, is the standard deviation of the risk value in the next node vector, is the report density in the latter node vector, is the number of reports in the latter node vector, is the average risk value in the previous node vector, is the standard deviation of the risk value in the previous node vector, is the report density in the previous node vector, is the number of reports in the previous node vector;

[0023] Furthermore, the comprehensive prediction unit includes a data comprehensive processor and a probability prediction processor, wherein the data comprehensive processor is used to comprehensively process the step vector and the time vector, and the probability prediction processor calculates the probability of occurrence of the public opinion accident based on the processing result;

[0024] The data integration processor processes the step vector and the time vector according to the following formula:

[0025] ;

[0026] in, represents the value of the i-th element in the ladder vector, represents the i-th element value in the time vector, T0 represents the time base, and P is the comprehensive judgment value;

[0027] The probability prediction processor calculates the probability Q of a public opinion accident according to the following formula:

[0028] ;

[0029] Among them, P0 is the standard judgment value.

[0030] The beneficial effects achieved by the present invention are:

[0031] This system processes text based on AI technology, and then obtains graph information after quantification. It analyzes the evolution of public opinion based on the graph information, and combines the risk characteristics of graph nodes and the time correlation characteristics between graph nodes for analysis. It can effectively explore the development characteristics of public opinion and predict more accurate probability information based on this.

[0032] To further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are only for reference and description and are not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the overall structural framework of the present invention;

[0034] Figure 2 A schematic diagram of the acquisition module structure is provided for the purpose of the present invention;

[0035] Figure 3 This is a schematic diagram of the text parsing processing module of the present invention;

[0036] Figure 4 This is a schematic diagram of the event graph analysis module of the present invention;

[0037] Figure 5 This is a schematic diagram of the risk model prediction module of the present invention. DETAILED DESCRIPTION

[0038] The following is an explanation of the embodiments of the present invention through specific embodiments. Those skilled in the art can understand the advantages and effects of the present invention from the contents disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and the details in this specification can also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. In addition, the drawings of the present invention are only simple schematic illustrations and are not depicted according to actual sizes. It is stated in advance. The following embodiments will further explain the relevant technical contents of the present invention in detail, but the disclosed contents are not intended to limit the scope of protection of the present invention.

[0039] Embodiment 1: This embodiment provides a public opinion risk evolution prediction system based on AI technology, combined with Figure 1 , including target setting acquisition module, text parsing and processing module, event graph parsing module and risk model prediction module;

[0040] The target setting collection module is used to input the target event and collect network public opinion information related to the target event, the text analysis processing module is used to analyze and obtain the risk points in the public opinion information, the event map analysis module is used to construct a development map of the risk points, and the risk model prediction module uses the development map information as input to perform risk prediction on the target event;

[0041] The text analysis processing module includes a keyword extraction unit, a risk point matching unit and a risk quantification unit. The keyword extraction unit is used to extract keyword information in the report content, the risk point matching unit is used to match the keyword with the historical case information, and the risk quantification unit is used to calculate the risk value of the report content;

[0042] The event graph analysis module includes an event normalization unit, a risk feature unit and a time management unit. The event normalization unit is used to unify the report content describing the same event into a graph node. The risk feature unit is used to analyze and obtain the risk feature of each graph node. The time management unit is used to identify and manage the time relationship between graph nodes.

[0043] The risk model prediction module includes a graph input unit, a step processing unit and a comprehensive prediction unit. The graph input unit is used to input graph information, the step processing unit is used to analyze and process the development trend between two adjacent graph nodes, and the comprehensive prediction unit is used to generate probability information of the final public opinion accident.

[0044] The risk quantification unit includes a weight calculation processor and a risk calculation processor, wherein the weight calculation processor is used to calculate the risk weight coefficient of each keyword in the report content, and the risk calculation processor is used to calculate the risk value of each report content;

[0045] The weight calculation processor calculates the risk weight coefficient k of the keyword according to the following formula:

[0046] ;

[0047] in, is the number of matching cases containing the corresponding keyword, and m is the number of matching cases;

[0048] The risk calculation processor calculates the risk value Vr of each report content according to the following formula:

[0049] ;

[0050] Among them, N is the number of keywords contained in the report content, k i represents the risk weight coefficient of the ith keyword in the report content;

[0051] The ladder processing unit includes a vector management processor and a ladder calculation processor, wherein the vector management processor is used to receive a node vector and send two adjacent node vectors to the ladder calculation processor, and the ladder calculation processor is used to calculate the ladder change values ​​of the two node vectors;

[0052] The step calculation processor calculates the step change value according to the following formula: :

[0053] ;

[0054] in, is the average risk value in the next node vector, is the standard deviation of the risk value in the next node vector, is the report density in the latter node vector, is the number of reports in the latter node vector, is the average risk value in the previous node vector, is the standard deviation of the risk value in the previous node vector, is the report density in the previous node vector, is the number of reports in the previous node vector;

[0055] The comprehensive prediction unit includes a data comprehensive processor and a probability prediction processor, wherein the data comprehensive processor is used to comprehensively process the step vector and the time vector, and the probability prediction processor calculates the probability of occurrence of the public opinion accident based on the processing result;

[0056] The data integration processor processes the step vector and the time vector according to the following formula:

[0057] ;

[0058] in, represents the value of the i-th element in the ladder vector, represents the i-th element value in the time vector, T0 represents the time base, and P is the comprehensive judgment value;

[0059] The probability prediction processor calculates the probability Q of a public opinion accident according to the following formula:

[0060] ;

[0061] Among them, P0 is the standard judgment value.

[0062] Embodiment 2: This embodiment includes all the contents of Embodiment 1, and provides a public opinion risk evolution prediction system based on AI technology, including a target setting acquisition module, a text parsing processing module, an event graph parsing module, and a risk model prediction module;

[0063] The target setting collection module is used to input the target event and collect network public opinion information related to the target event, the text analysis processing module is used to analyze and obtain the risk points in the public opinion information, the event map analysis module is used to construct a development map of the risk points, and the risk model prediction module uses the development map information as input to perform risk prediction on the target event;

[0064] Combination Figure 2 The target setting acquisition module includes a target input unit, a report retrieval unit and a text storage unit, wherein the target input unit is used to input simple description information of the target event, the report retrieval unit is used to retrieve report content associated with the simple description information on the Internet, and the text storage unit is used to store the retrieved report content;

[0065] Combination Figure 3 The text analysis processing module includes a keyword extraction unit, a risk point matching unit and a risk quantification unit. The keyword extraction unit is used to extract keyword information in the report content, the risk point matching unit is used to match the keyword with the historical case information, and the risk quantification unit is used to calculate the risk value of the report content;

[0066] Combination Figure 4 The event graph analysis module includes an event normalization unit, a risk feature unit and a time management unit. The event normalization unit is used to unify the report content describing the same event into a graph node, the risk feature unit is used to analyze and obtain the risk feature of each graph node, and the time management unit is used to identify and manage the time relationship between graph nodes;

[0067] Combination Figure 5 The risk model prediction module includes a graph input unit, a step processing unit and a comprehensive prediction unit. The graph input unit is used to input graph information, the step processing unit is used to analyze and process the development trend between two adjacent graph nodes, and the comprehensive prediction unit is used to make probability information of the final public opinion accident;

[0068] The target input unit includes a window display processor and an information receiving processor, wherein the target input unit is used to display a window for inputting information, and the information receiving processor is used to receive input information of the window;

[0069] The report retrieval unit includes a correlation retrieval processor, a time screening processor and an information copy processor, wherein the correlation retrieval processor is used to retrieve the report content containing the target event information on the Internet, the time screening processor is used to screen the recent report content, and the information copy processor is used to copy the screened report content;

[0070] The text storage unit includes an information register and a timing management processor, wherein the information register is used to store text information of the report content, and the timing management processor controls and manages the storage order of the text information based on the report time;

[0071] The keyword extraction unit includes a keyword register and a comparison extraction processor, wherein the keyword register is used to store preset keyword information, and the comparison extraction processor is used to compare the text information with the information in the keyword register to extract the keywords in the text information;

[0072] The risk point matching unit includes a case search processor, a similarity calculation processor and a case matching processor, wherein the case search processor is used to find case information containing the same keywords, the similarity calculation processor is used to calculate the similarity index between the case information and the report content, and the case matching processor selects matching case information based on the similarity index;

[0073] The similarity calculation processor calculates the similarity index Sp between the case information and the report content according to the following formula:

[0074] ;

[0075] Among them, A i represents the i-th element value of the keyword vector of the report content, B i represents the i-th element value of the historical case keyword vector, and n is the number of elements in the keyword vector;

[0076] The value of an element in the keyword vector is 1 or 0, 1 indicates that the corresponding keyword is contained, and 0 indicates that the corresponding keyword is not contained. The number of elements in the keyword vector is equal to the number of keywords in the keyword register;

[0077] The case matching processor selects m historical cases with the largest similarity index as matching cases;

[0078] The risk quantification unit includes a weight calculation processor and a risk calculation processor, wherein the weight calculation processor is used to calculate the risk weight coefficient of each keyword in the report content, and the risk calculation processor is used to calculate the risk value of each report content;

[0079] The weight calculation processor calculates the risk weight coefficient k of the keyword according to the following formula:

[0080] ;

[0081] in, is the number of matching cases containing the corresponding keywords;

[0082] The risk calculation processor calculates the risk value Vr of each report content according to the following formula:

[0083] ;

[0084] Among them, N is the number of keywords contained in the report content, k i represents the risk weight coefficient of the ith keyword in the report content;

[0085] The event normalization unit includes a subject identification processor and a node information register, wherein the subject identification processor is used to identify the subject information of each report content and treat the report content with the same subject information as a graph node, and the node information register is used to store the risk value data of all report contents in each graph node;

[0086] The risk feature unit includes a fluctuation feature processor and a mean feature processor, wherein the fluctuation feature processor is used to calculate the standard deviation of the risk value of each graph node, and the mean feature processor is used to calculate the average value of the risk value of each graph node;

[0087] The time management unit includes a starting point setting processor, an interval setting processor and a density setting processor. The starting point setting processor is used to set the starting time of each graph node. The interval setting processor is used to calculate the time difference between the starting times of adjacent graph nodes as the interval time of the graph nodes. The density setting processor is used to calculate the coverage density of each graph node.

[0088] The earliest publishing time of the reported content in each graph node is the starting time, and the density setting processor calculates the report density d of the graph node according to the following formula:

[0089] ;

[0090] Among them, nR is the number of reported contents in the graph node, and T is the time span of the release time of the reported contents in the graph node;

[0091] The graph input unit includes an input sorting processor and a dedicated transmission processor, wherein the input sorting processor is used to sort the graph information into a plurality of input items, and the dedicated transmission processor transmits the input items to a dedicated processing unit;

[0092] The input items include M node vectors and a time vector. Each node vector includes four elements: risk value standard deviation, risk value average, report density and report quantity. The time vector is a vector composed of interval time, including M-1 elements, where M is the number of graph nodes.

[0093] The node vector is sent to the ladder processing unit, and the time vector is sent to the comprehensive prediction unit;

[0094] The ladder processing unit includes a vector management processor and a ladder calculation processor, wherein the vector management processor is used to receive a node vector and send two adjacent node vectors to the ladder calculation processor, and the ladder calculation processor is used to calculate the ladder change values ​​of the two node vectors;

[0095] The step calculation processor calculates the step change value according to the following formula: :

[0096] ;

[0097] in, is the average risk value in the next node vector, is the standard deviation of the risk value in the next node vector, is the report density in the latter node vector, is the number of reports in the latter node vector, is the average risk value in the previous node vector, is the standard deviation of the risk value in the previous node vector, is the report density in the previous node vector, is the number of reports in the previous node vector;

[0098] The step calculation processor forms a step vector with M-1 step change values ​​and sends it to the comprehensive prediction unit;

[0099] The comprehensive prediction unit includes a data comprehensive processor and a probability prediction processor, wherein the data comprehensive processor is used to comprehensively process the step vector and the time vector, and the probability prediction processor calculates the probability of occurrence of the public opinion accident based on the processing result;

[0100] The data integration processor processes the step vector and the time vector according to the following formula:

[0101] ;

[0102] in, represents the value of the i-th element in the ladder vector, represents the i-th element value in the time vector, T0 represents the time base, and P is the comprehensive judgment value;

[0103] The probability prediction processor calculates the probability Q of a public opinion accident according to the following formula:

[0104] ;

[0105] Among them, P0 is the standard judgment value;

[0106] The i appearing in the above text is an ordinal number used to indicate a sequence number.

[0107] The contents disclosed above are only preferred feasible embodiments of the present invention, and do not limit the protection scope of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention specification and drawings are included in the protection scope of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. A public opinion risk evolution prediction system based on AI technology, characterized by: It includes target setting acquisition module, text analysis and processing module, event graph analysis module and risk model prediction module; The target setting collection module is used to input the target event and collect network public opinion information related to the target event, the text analysis processing module is used to analyze and obtain the risk points in the public opinion information, the event map analysis module is used to construct a development map of the risk points, and the risk model prediction module uses the development map information as input to perform risk prediction on the target event; The text analysis processing module includes a keyword extraction unit, a risk point matching unit and a risk quantification unit. The keyword extraction unit is used to extract keyword information in the report content, the risk point matching unit is used to match the keyword with the historical case information, and the risk quantification unit is used to calculate the risk value of the report content; The event graph analysis module includes an event normalization unit, a risk feature unit and a time management unit. The event normalization unit is used to unify the report content describing the same event into a graph node. The risk feature unit is used to analyze and obtain the risk feature of each graph node. The time management unit is used to identify and manage the time relationship between graph nodes. The risk model prediction module includes a graph input unit, a step processing unit and a comprehensive prediction unit. The graph input unit is used to input graph information, the step processing unit is used to analyze and process the development trend between two adjacent graph nodes, and the comprehensive prediction unit is used to generate probability information of the final public opinion accident. The risk quantification unit includes a weight calculation processor and a risk calculation processor, wherein the weight calculation processor is used to calculate the risk weight coefficient of each keyword in the report content, and the risk calculation processor is used to calculate the risk value of each report content; The weight calculation processor calculates the risk weight coefficient k of the keyword according to the following formula: ； in, is the number of matching cases containing the corresponding keyword, and m is the number of matching cases; The risk calculation processor calculates the risk value Vr of each report content according to the following formula: ； Among them, N is the number of keywords contained in the report content, k i represents the risk weight coefficient of the ith keyword in the report content; The ladder processing unit includes a vector management processor and a ladder calculation processor, wherein the vector management processor is used to receive a node vector and send two adjacent node vectors to the ladder calculation processor, and the ladder calculation processor is used to calculate the ladder change values ​​of the two node vectors; The step calculation processor calculates the step change value according to the following formula: : ； in, is the average risk value in the next node vector, is the standard deviation of the risk value in the next node vector, is the report density in the latter node vector, is the number of reports in the latter node vector, is the average risk value in the previous node vector, is the standard deviation of the risk value in the previous node vector, is the report density in the previous node vector, is the number of reports in the previous node vector; The comprehensive prediction unit includes a data comprehensive processor and a probability prediction processor, wherein the data comprehensive processor is used to comprehensively process the step vector and the time vector, and the probability prediction processor calculates the probability of occurrence of the public opinion accident based on the processing result; The data integration processor processes the step vector and the time vector according to the following formula: ； in, represents the value of the i-th element in the ladder vector, represents the i-th element value in the time vector, T0 represents the time base, P represents the comprehensive judgment value; M represents the number of graph nodes; The probability prediction processor calculates the probability Q of a public opinion accident according to the following formula: ； Among them, P0 is the standard judgment value.

Images