A public opinion risk determination system based on AI technology

Abstract

The present invention provides a public opinion risk determination system based on AI technology, which relates to the field of electronic digital data processing, including a data collection module, a text processing module, a model construction module and a comprehensive rating module. The data collection module is used to collect public opinion information, the text processing module is used to perform text analysis on the collected public opinion information, the model construction module is used to organize and count the analysis results to obtain a public opinion model of the event, and the comprehensive evaluation module is used to perform a comprehensive analysis on the public opinion model to determine risk information; this system can monitor and analyze public opinion data in real time, and determine public opinion risks in a timely manner through a multi-level and multi-dimensional analysis model.

Description

Technical Field

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

[0002] With the popularization of the Internet, more and more people post their views on public events on social platforms. If public opinion is not guided correctly, it is easy to generate public opinion risks and aggravate social conflicts and divisions. Therefore, there is a need for a system that can automatically monitor public opinion and promptly detect public opinion risks so that relevant parties can prepare response measures to public opinion in advance.

[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 analysis systems have been developed. After a lot of searching and reference, it is found that the existing public opinion analysis systems are such as the system disclosed by the publication number CN112395539B. These system methods generally include step 1, obtaining public opinion information on the Internet, using the text extraction model to extract the data structured content of the public opinion information, and obtaining the data structured content; step 2, using the document theme generation model to generate the document theme of the public opinion information; step 3, using the short text position determination model based on natural language processing to determine the position of the public opinion information; step 4, using the public opinion propagation network traceability analysis model to trace the public opinion information, and establish a time-series network graph structure snapshot of the user interaction behavior according to a certain time interval and the traceability analysis results. The network graph structure snapshot includes data structured content, document theme and position determination. However, the system only traces the public opinion information, and does not analyze and judge the public opinion risks generated, and cannot help the relevant subjects to make further preparations and response measures. Summary of the invention

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

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

[0007] A public opinion risk determination system based on AI technology, including a data collection module, a text processing module, a model building module and a comprehensive rating module;

[0008] The data collection module is used to collect public opinion information, the text processing module is used to perform text analysis on the collected public opinion information, the model building module is used to organize and compile statistics on the analysis results to obtain a public opinion model of the event, and the comprehensive evaluation module is used to perform a comprehensive analysis on the public opinion model to determine risk information;

[0009] The data collection module includes a platform docking unit, a data screening unit and a text storage unit, wherein the platform docking unit is used to connect to the social platform, the data screening unit is used to screen out the public opinion text information, and the text storage unit is used to store the screened public opinion text information;

[0010] The text processing module includes a short sentence disassembly unit, a keyword matching unit and a sentiment analysis unit. The short sentence disassembly unit is used to disassemble each public opinion text information into multiple short sentences. The keyword matching unit is used to match the keywords in the short sentences. The sentiment analysis unit analyzes the matching results to obtain the sentiment evaluation value of each public opinion text information.

[0011] The model building module includes a node framework unit and a node feature unit, wherein the node framework unit is used to build a model framework and deploy event nodes to which the sentiment evaluation value is attached, and the node feature unit is used to calculate the public opinion feature of each event node;

[0012] The comprehensive rating module includes a weight evaluation unit, a source correction unit and a risk evaluation unit. The weight evaluation unit obtains the weight value of each event based on the discussion scale of the event, the source correction unit corrects the weight value based on the information source of the event, and the risk evaluation unit is used to perform risk evaluation on the public opinion model;

[0013] Furthermore, the sentiment analysis unit includes a sentiment classification processor and a sentiment evaluation processor, wherein the sentiment classification processor is used to classify short sentences based on sentiment color values, and the sentiment evaluation processor is used to calculate the sentiment evaluation value of each comment type text information;

[0014] The sentiment classification processor classifies short sentences into positive sentiment classes, neutral sentiment classes, and negative sentiment classes;

[0015] The sentiment evaluation processor calculates the sentiment evaluation value Ve of each comment type text information according to the following formula:

[0016] ;

[0017] Among them, n1 is the number of short sentences with positive emotions, n2 is the number of short sentences with neutral emotions, and n3 is the number of short sentences with negative emotions;

[0018] When Ve is greater than 1, set Ve to 1, and when Ve is less than 0, set Ve to 0;

[0019] Further, the node feature unit includes a mean calculation processor, a standard deviation calculation processor and a scale calculation processor, wherein the mean calculation processor is used to calculate the mean of the sentiment evaluation value of each event node, the standard deviation calculation processor is used to calculate the standard deviation of the sentiment evaluation value of each event node, and the scale calculation processor is used to calculate the discussion scale index of each event node;

[0020] Furthermore, the weight comparison processor calculates the reference value W0 according to the following formula:

[0021] ;

[0022] Where m is the number of event nodes in the model, and Ds(i) represents the number of discussion scales of the i-th event node;

[0023] The weight calculation processor calculates the weight coefficient k(i) of the i-th event node according to the following formula:

[0024] ;

[0025] in, is the relative coefficient;

[0026] Further, the risk assessment unit includes a risk calculation processor and a feedback warning processor, the risk calculation processor is used to calculate the public opinion risk index of each model, and the feedback warning processor makes feedback based on the public opinion risk index;

[0027] The risk calculation processor calculates the public opinion risk index Q according to the following formula:

[0028] ;

[0029] in, represents the mean value of the sentiment evaluation value of the i-th event node, Represents the standard deviation of the sentiment evaluation value of the i-th event node.

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

[0031] This system ensures comprehensive coverage of public opinion data through multi-platform data collection, performs short sentence splitting and keyword matching on reporting documents to achieve refined public opinion analysis, uses AI technology to perform sentiment analysis and risk assessment to improve the accuracy of public opinion analysis, and provides multi-dimensional comprehensive assessments based on source ratings and risk types to ensure the comprehensiveness and reliability of public opinion risk assessments. It collects and analyzes data in real time, generates early warning information in a timely manner, and helps users respond quickly to public opinion risks.

[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 This is a schematic diagram of the data collection module of the present invention;

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

[0036] Figure 4 This is a schematic diagram of the module structure of the model of the present invention;

[0037] Figure 5 This is a schematic diagram of the comprehensive rating 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 determination system based on AI technology, combined with Figure 1 , including data collection module, text processing module, model building module and comprehensive rating module;

[0040] The data collection module is used to collect public opinion information, the text processing module is used to perform text analysis on the collected public opinion information, the model building module is used to organize and compile statistics on the analysis results to obtain a public opinion model of the event, and the comprehensive evaluation module is used to perform a comprehensive analysis on the public opinion model to determine risk information;

[0041] The data collection module includes a platform docking unit, a data screening unit and a text storage unit, wherein the platform docking unit is used to connect to the social platform, the data screening unit is used to screen out the public opinion text information, and the text storage unit is used to store the screened public opinion text information;

[0042] The text processing module includes a short sentence disassembly unit, a keyword matching unit and a sentiment analysis unit. The short sentence disassembly unit is used to disassemble each public opinion text information into multiple short sentences. The keyword matching unit is used to match the keywords in the short sentences. The sentiment analysis unit analyzes the matching results to obtain the sentiment evaluation value of each public opinion text information.

[0043] The model building module includes a node framework unit and a node feature unit, wherein the node framework unit is used to build a model framework and deploy event nodes to which the sentiment evaluation value is attached, and the node feature unit is used to calculate the public opinion feature of each event node;

[0044] The comprehensive rating module includes a weight evaluation unit, a source correction unit and a risk evaluation unit. The weight evaluation unit obtains the weight value of each event based on the discussion scale of the event, the source correction unit corrects the weight value based on the information source of the event, and the risk evaluation unit is used to perform risk evaluation on the public opinion model;

[0045] The sentiment analysis unit includes a sentiment classification processor and a sentiment evaluation processor, wherein the sentiment classification processor is used to classify short sentences based on sentiment color values, and the sentiment evaluation processor is used to calculate the sentiment evaluation value of each comment type text information;

[0046] The sentiment classification processor classifies short sentences into positive sentiment classes, neutral sentiment classes, and negative sentiment classes;

[0047] The sentiment evaluation processor calculates the sentiment evaluation value Ve of each comment type text information according to the following formula:

[0048] ;

[0049] Among them, n1 is the number of short sentences with positive emotions, n2 is the number of short sentences with neutral emotions, and n3 is the number of short sentences with negative emotions;

[0050] When Ve is greater than 1, set Ve to 1, and when Ve is less than 0, set Ve to 0;

[0051] The node feature unit includes a mean calculation processor, a standard deviation calculation processor and a scale calculation processor, wherein the mean calculation processor is used to calculate the mean of the sentiment evaluation value of each event node, the standard deviation calculation processor is used to calculate the standard deviation of the sentiment evaluation value of each event node, and the scale calculation processor is used to calculate the discussion scale index of each event node;

[0052] The weight comparison processor calculates the reference value W0 according to the following formula:

[0053] ;

[0054] Where m is the number of event nodes in the model, and Ds(i) represents the number of discussion scales of the i-th event node;

[0055] The weight calculation processor calculates the weight coefficient k(i) of the i-th event node according to the following formula:

[0056] ;

[0057] in, is the relative coefficient;

[0058] The risk assessment unit includes a risk calculation processor and a feedback warning processor, wherein the risk calculation processor is used to calculate the public opinion risk index of each model, and the feedback warning processor makes feedback based on the public opinion risk index;

[0059] The risk calculation processor calculates the public opinion risk index Q according to the following formula:

[0060] ;

[0061] in, represents the mean value of the sentiment evaluation value of the i-th event node, Represents the standard deviation of the sentiment evaluation value of the i-th event node.

[0062] Embodiment 2: This embodiment includes all the contents of Embodiment 1, and provides a public opinion risk determination system based on AI technology, including a data collection module, a text processing module, a model building module and a comprehensive rating module;

[0063] The data collection module is used to collect public opinion information, the text processing module is used to perform text analysis on the collected public opinion information, the model building module is used to organize and compile statistics on the analysis results to obtain a public opinion model of the event, and the comprehensive evaluation module is used to perform a comprehensive analysis on the public opinion model to determine risk information;

[0064] Combination Figure 2 The data collection module includes a platform docking unit, a data screening unit and a text storage unit, wherein the platform docking unit is used to connect to the social platform, the data screening unit is used to screen out the public opinion text information, and the text storage unit is used to store the screened public opinion text information;

[0065] Combination Figure 3The text processing module includes a short sentence disassembly unit, a keyword matching unit and a sentiment analysis unit. The short sentence disassembly unit is used to disassemble each public opinion text information into multiple short sentences. The keyword matching unit is used to match the keywords in the short sentences. The sentiment analysis unit analyzes the matching results to obtain the sentiment evaluation value of each public opinion text information.

[0066] Combination Figure 4 , the model building module includes a node framework unit and a node feature unit, the node framework unit is used to build a model framework and deploy event nodes to which the sentiment evaluation value is attached, and the node feature unit is used to calculate the public opinion feature of each event node;

[0067] The public opinion model is composed of event nodes, which are divided into basic event nodes and extended event nodes. Each public opinion model has only one basic event node, and the number of extended event nodes in each public opinion model increases continuously according to the development of public opinion.

[0068] Combination Figure 5 The comprehensive rating module includes a weight evaluation unit, a source correction unit and a risk evaluation unit. The weight evaluation unit obtains the weight value of each event based on the discussion scale of the event, the source correction unit corrects the weight value based on the information source of the event, and the risk evaluation unit is used to perform risk evaluation on the public opinion model;

[0069] The platform docking unit includes an API interface management processor, a crawler scheduling processor and a data format conversion processor. The API interface management processor is used to store and connect the API interface of the social platform, the crawler scheduling processor is used to crawl data from the connected social platform, and the data format conversion processor is used to convert the crawled data into the same text format;

[0070] The data screening unit includes a time screening processor and a type screening processor, wherein the time screening processor is used to screen out text content within the current period, and the type screening processor is used to screen out text content of the public opinion type;

[0071] The public opinion types of public opinion information are divided into event type and comment type. The event type is a description of an event, and the comment type is a comment on the event.

[0072] The text storage unit includes an event storage, a comment storage and an association processor, wherein the event storage is used to store text information of event type, the comment storage is used to store text information of comment type, and the association processor is used to record the dependency relationship between the text information of comment type and the text information of event type;

[0073] The short sentence disassembling unit includes a word segmentation processor, a grammar analysis processor, a short sentence generation processor and a tag processor, wherein the word segmentation processor is used to extract independent words from text information, the grammar analysis processor is used to identify the part of speech of the words and group the words based on grammar, the short sentence generation processor is used to synthesize the grouped words into a short sentence, and the tag processor is used to set an attribute tag for each short sentence;

[0074] The keyword matching unit includes a keyword extraction processor and a color matching processor, wherein the keyword extraction processor is used to extract keywords from short sentences, and the color matching processor obtains corresponding emotional color values ​​based on the extracted keyword matching;

[0075] The sentiment analysis unit includes a sentiment classification processor and a sentiment evaluation processor, wherein the sentiment classification processor is used to classify short sentences based on sentiment color values, and the sentiment evaluation processor is used to calculate the sentiment evaluation value of each comment type text information;

[0076] The sentiment classification processor classifies short sentences into positive sentiment classes, neutral sentiment classes, and negative sentiment classes;

[0077] The sentiment evaluation processor calculates the sentiment evaluation value Ve of each comment type text information according to the following formula:

[0078] ;

[0079] Among them, n1 is the number of short sentences with positive emotions, n2 is the number of short sentences with neutral emotions, and n3 is the number of short sentences with negative emotions;

[0080] When Ve is greater than 1, set Ve to 1, and when Ve is less than 0, set Ve to 0;

[0081] The node framework unit includes a node mapping processor, a node concentration processor and a data deployment processor. The node mapping processor maps event type text information to event nodes. The node concentration processor forms a set of associated event nodes and uses the first event node in the set as the basic event node, and the remaining event nodes in the set as extended event nodes. The data deployment processor creates a storage space based on each event node for storing the corresponding emotion evaluation value.

[0082] The node feature unit includes a mean calculation processor, a standard deviation calculation processor and a scale calculation processor, wherein the mean calculation processor is used to calculate the mean of the sentiment evaluation value of each event node, the standard deviation calculation processor is used to calculate the standard deviation of the sentiment evaluation value of each event node, and the scale calculation processor is used to calculate the discussion scale index of each event node;

[0083] The scale calculation processor calculates the discussion scale index Ds according to the following formula:

[0084] ;

[0085] Where N is the number of sentiment evaluation values ​​for each event node;

[0086] The weight evaluation unit includes a weight comparison processor and a weight calculation processor, wherein the weight comparison processor is used to compare the discussion scale index of event nodes under the same model to calculate a benchmark value, and the weight calculation processor calculates a weight coefficient of each event node based on the benchmark value;

[0087] The weight comparison processor calculates the reference value W0 according to the following formula:

[0088] ;

[0089] Where m is the number of event nodes in the model, and Ds(i) represents the number of discussion scales of the i-th event node;

[0090] The weight calculation processor calculates the weight coefficient k(i) of the i-th event node according to the following formula:

[0091] ;

[0092] in, is the relative coefficient;

[0093] The source correction unit includes a source retrieval processor, a platform parameter register and a platform correction processor. The source retrieval processor is used to retrieve the source platform of each comment type text information and determine the main source platform of each event node. The platform parameter register is used to store the correction parameters of each social platform. The platform correction processor performs correction calculation on the weight coefficient of the event node based on the correction parameters of the main source platform.

[0094] The platform correction processor calculates the corrected weight coefficient according to the following formula: :

[0095] ;

[0096] in, is the influence parameter of the social platform, It is the credibility parameter of the social platform;

[0097] The risk assessment unit includes a risk calculation processor and a feedback warning processor, wherein the risk calculation processor is used to calculate the public opinion risk index of each model, and the feedback warning processor makes feedback based on the public opinion risk index;

[0098] The risk calculation processor calculates the public opinion risk index Q according to the following formula:

[0099] ;

[0100] in, represents the mean value of the sentiment evaluation value of the i-th event node, Represents the standard deviation of the sentiment evaluation value of the i-th event node;

[0101] When the public opinion risk index is greater than a threshold, the feedback warning processor issues a public opinion warning;

[0102] The i appearing in the above text is an ordinal number used to indicate a sequence number.

[0103] 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 determination system based on AI technology, characterized in that: It includes data collection module, text processing module, model building module and comprehensive rating module; The data collection module is used to collect public opinion information, the text processing module is used to perform text analysis on the collected public opinion information, the model building module is used to organize and compile statistics on the analysis results to obtain a public opinion model of the event, and the comprehensive rating module is used to perform a comprehensive analysis on the public opinion model to determine risk information; The data collection module includes a platform docking unit, a data screening unit and a text storage unit, wherein the platform docking unit is used to connect to the social platform, the data screening unit is used to screen out the public opinion text information, and the text storage unit is used to store the screened public opinion text information; The text processing module includes a short sentence disassembly unit, a keyword matching unit and a sentiment analysis unit. The short sentence disassembly unit is used to disassemble each public opinion text information into multiple short sentences. The keyword matching unit is used to match the keywords in the short sentences. The sentiment analysis unit analyzes the matching results to obtain the sentiment evaluation value of each public opinion text information. The model building module includes a node framework unit and a node feature unit, wherein the node framework unit is used to build a model framework and deploy event nodes to which the sentiment evaluation value is attached, and the node feature unit is used to calculate the public opinion feature of each event node; The node framework unit includes a node mapping processor, a node concentration processor and a data deployment processor. The node mapping processor maps event type text information to event nodes. The node concentration processor forms a set of associated event nodes and uses the first event node in the set as the basic event node, and the remaining event nodes in the set as extended event nodes. The data deployment processor creates a storage space based on each event node for storing the corresponding emotion evaluation value. The data screening unit includes a time screening processor and a type screening processor, wherein the time screening processor is used to screen out text content within the current period, and the type screening processor is used to screen out text content of the public opinion type; The public opinion types of public opinion information are divided into event type and comment type. The event type is a description of an event, and the comment type is a comment on the event. The comprehensive rating module includes a weight evaluation unit, a source correction unit and a risk evaluation unit. The weight evaluation unit obtains the weight value of each event based on the discussion scale of the event, the source correction unit corrects the weight value based on the information source of the event, and the risk evaluation unit is used to perform risk evaluation on the public opinion model; The public opinion model is composed of event nodes, which are divided into basic event nodes and extended event nodes. Each public opinion model has only one basic event node, and the number of extended event nodes in each public opinion model increases continuously according to the development of public opinion. The node feature unit includes a mean calculation processor, a standard deviation calculation processor and a scale calculation processor, wherein the mean calculation processor is used to calculate the mean of the sentiment evaluation value of each event node, the standard deviation calculation processor is used to calculate the standard deviation of the sentiment evaluation value of each event node, and the scale calculation processor is used to calculate the discussion scale index of each event node; The scale calculation processor calculates the discussion scale index Ds according to the following formula: ； Where N is the number of sentiment evaluation values ​​for each event node; The weight evaluation unit includes a weight comparison processor and a weight calculation processor, wherein the weight comparison processor is used to compare the discussion scale index of event nodes under the same model to calculate a benchmark value, and the weight calculation processor calculates a weight coefficient of each event node based on the benchmark value; The weight comparison processor calculates the reference value W0 according to the following formula: ； Where m is the number of event nodes in the model, and Ds(i) represents the number of discussion scales of the i-th event node; The weight calculation processor calculates the weight coefficient k(i) of the i-th event node according to the following formula: ； in, is the relative coefficient; The source correction unit includes a source retrieval processor, a platform parameter register and a platform correction processor. The source retrieval processor is used to retrieve the source platform of each comment type text information and determine the main source platform of each event node. The platform parameter register is used to store the correction parameters of each social platform. The platform correction processor performs correction calculation on the weight coefficient of the event node based on the correction parameters of the main source platform. The platform correction processor calculates the corrected weight coefficient according to the following formula: : ； in, is the influence parameter of the social platform, It is the credibility parameter of the social platform; The risk assessment unit includes a risk calculation processor and a feedback warning processor, wherein the risk calculation processor is used to calculate the public opinion risk index of each model, and the feedback warning processor makes feedback based on the public opinion risk index; The risk calculation processor calculates the public opinion risk index Q according to the following formula: ； in, represents the mean value of the sentiment evaluation value of the i-th event node, Represents the standard deviation of the sentiment evaluation value of the i-th event node; The sentiment analysis unit includes a sentiment classification processor and a sentiment evaluation processor, wherein the sentiment classification processor is used to classify short sentences based on sentiment color values, and the sentiment evaluation processor is used to calculate the sentiment evaluation value of each comment type text information; The sentiment classification processor classifies short sentences into positive sentiment classes, neutral sentiment classes, and negative sentiment classes; The sentiment evaluation processor calculates the sentiment evaluation value Ve of each comment type text information according to the following formula: ； Among them, n1 is the number of short sentences with positive emotions, n2 is the number of short sentences with neutral emotions, and n3 is the number of short sentences with negative emotions; When Ve is greater than 1, set Ve to 1; when Ve is less than 0, set Ve to 0.

Images