Public opinion means intelligent identification system based on artificial intelligence large model

Abstract

The invention relates to the field of public opinion means intelligent identification, in particular to a public opinion means intelligent identification system based on an artificial intelligence large model, and the reliability and accuracy of an identification result are improved. In the system, a public opinion data storage module adopts a distributed database to store original public opinion data, a knowledge base storage module adopts a graph database to store a public opinion and tactical knowledge system, and a result cache module adopts a high-speed cache to store an analysis result; public opinion event data are acquired through the data acquisition preprocessing module and are cleaned and standardized, after the data are standardized, core element information of public opinion events is identified and extracted through the event element extraction module, and then real development venation of the events is checked and confirmed through the fact verification module. And then intelligent identification of public opinion means is carried out through the public opinion and tactical method identification module, and finally the identification result is confirmed and corrected through the man-machine collaborative verification module. The method is suitable for public opinion identification.

Description

Technical Field

[0001] The present invention relates to the field of intelligent identification of public opinion means, and particularly relates to an intelligent identification system of public opinion means based on an artificial intelligence large model. Background Art

[0002] At present, traditional public opinion analysis systems mainly focus on surface features such as the emotional tendency, dissemination trend, and hot topics of information, and have formed relatively mature analysis frameworks and technical methods. These systems usually adopt technologies such as machine learning and natural language processing to achieve automated processing and analysis of massive public opinion data.

[0003] However, in a complex public opinion environment, simply relying on basic functions such as sentiment analysis and hot topic discovery has been difficult to meet the needs of in-depth public opinion analysis. Especially in identifying and coping with various public opinion warfare methods, traditional technologies have obvious limitations.

[0004] The existing public opinion analysis technologies mainly have the following technical defects:

[0005] Single analysis dimension: Existing public opinion analysis systems mainly focus on surface features such as the emotional polarity and dissemination volume of texts, and it is difficult to deeply identify and understand complex public opinion strategies and discourse methods. Specifically manifested as: only being able to identify basic emotional tendencies such as positive, negative, and neutral of texts, unable to understand the implicit dissemination intentions and strategies in texts, and lacking the ability to analyze the dissemination path and influence mechanism of public opinion;

[0006] Lack of context understanding: Existing machine learning models perform poorly in dealing with tasks that require in-depth context understanding, and it is difficult to accurately identify public opinion means such as framing and exaggeration that need to be combined with specific backgrounds. The main problems include: being unable to accurately understand the context correlation, having limited understanding ability for long texts, and being difficult to capture subtle semantic differences;

[0007] Difficulty in fact verification: Traditional systems lack the ability to compare and verify public opinion content with the factual basis, and are unable to effectively identify false information dissemination means such as "fabricating out of thin air". Specifically reflected in: lacking an automated fact verification mechanism, being unable to establish the mapping relationship between facts and public opinion, and having a low accuracy rate of authenticity judgment;

[0008] High labor cost: The existing identification of public opinion warfare methods mainly relies on manual judgment, which requires a large number of professional personnel to be invested, with low efficiency and difficulty in large-scale operation. The following problems exist: manual review takes a long time, with the average review time for a single piece of content exceeding 5 minutes, the training cost of professional personnel is high, and it is difficult to meet the rapid response requirements for sudden public opinion;

[0009] Poor interpretability: Traditional machine learning models often operate as black boxes when identifying public opinion warfare tactics, making it difficult to provide clear judgment bases and explanations. The main manifestations are as follows: unable to explain the model's decision-making process, lacking visual analysis evidence, and the credibility of the results being difficult to quantify;

[0010] Insufficient adaptability: Existing systems are difficult to quickly adapt to newly emerging public opinion warfare tactics and communication strategies, lacking sufficient flexibility and scalability. Specific problems include: long model update cycles, difficult knowledge base expansion, and poor recognition capabilities for new types of public opinion warfare tactics. Summary of the Invention

[0011] The purpose of the present invention is to overcome the shortcomings of the prior art and provide an intelligent recognition system for public opinion means based on an artificial intelligence large model, which improves the reliability and accuracy of recognition results.

[0012] The present invention adopts the following technical solutions to achieve the above purpose. The present invention provides an intelligent recognition system for public opinion means based on an artificial intelligence large model, including a data layer and a service layer. The data layer includes a public opinion data storage module, a knowledge base storage module, and a result cache module. The public opinion data storage module stores the original public opinion data using a distributed database. The knowledge base storage module stores the public opinion warfare knowledge system using a graph database. The result cache module stores the analysis results using a high-speed cache;

[0013] The service layer includes a data collection and preprocessing module, an event element extraction module, a fact verification module, a public opinion warfare recognition module, and a human-machine collaborative verification module. The public opinion event data is obtained through the data collection and preprocessing module, and is cleaned and standardized. After standardizing the data, the core element information of the public opinion event is identified and extracted through the event element extraction module. Then, the true development context of the event is verified and confirmed through the fact verification module. Next, intelligent recognition of public opinion means is carried out through the public opinion warfare recognition module. Finally, the recognition result is confirmed and corrected through the human-machine collaborative verification module.

[0014] Furthermore, the system also includes an interface layer. The interface layer includes a standard service interface, a real-time push interface, and an asynchronous task interface. The standard service interface is used to provide standardized service calls. The real-time push interface is used to support real-time data pushing. The asynchronous task interface is used to process asynchronous tasks.

[0015] Furthermore, the specific process of identifying and extracting the core element information of the public opinion event through the event element extraction module includes:

[0016] The basic theme and scope of the public opinion event to be analyzed are determined through the event element extraction module, and the key elements of the public opinion event are extracted, including time, place, person, and the course of the event.

[0017] Furthermore, the public opinion warfare method recognition module is specifically used to construct a public opinion warfare method knowledge base. The construction of the public opinion warfare method knowledge base specifically includes:

[0018] S201. Conduct a formal definition of public opinion warfare methods;

[0019] For the type of fanning the flames, the formal definition is as follows:

[0020] Basic characteristics: Exaggeration, rendering, or over-interpretation behaviors during the spread of public opinion events;

[0021] Core elements: Secondary spread, emotional rendering, and influence expansion;

[0022] Judgment dimensions: Temporal characteristics, content characteristics, and communication characteristics;

[0023] Quantitative index system:

[0024] Text deviation: Similarity threshold with the original fact;

[0025] Emotional intensity: Proportion of emotion words and polarity value threshold;

[0026] Communication influence: Topic growth rate and communication scope threshold;

[0027] Keyword characteristics: Frequency threshold of inciting words;

[0028] Comprehensive score: Multi-dimensional weighted calculation method;

[0029] For the type of framing, the formal definition is as follows:

[0030] Basic characteristics: Behaviors of fabricating or distorting facts and attributing them to specific objects;

[0031] Core elements: False information, causal distortion, and target orientation;

[0032] Judgment dimensions: Fact deviation, target orientation, and causal relationship;

[0033] Quantitative index system:

[0034] Fact consistency: Matching degree threshold with verified facts;

[0035] Target orientation: Frequency threshold of specific object mentions;

[0036] Causal credibility: Semantic relationship analysis threshold;

[0037] Comprehensive score: Multi-dimensional weighted calculation method;

[0038] S202. Construct a public opinion warfare method knowledge base;

[0039] Build a basic knowledge system, including the classification system of public opinion warfare methods, the formal definition specification standards, the calculation methods of quantitative indicators, and the rules for constructing evidence chains;

[0040] Case library construction, including the annotation specifications of typical cases, the case feature extraction methods, the case classification index mechanism, and the similar case matching rules;

[0041] Dynamic update mechanism, including the extraction of new warfare method features, the dynamic adjustment of the indicator system, the adaptive optimization of thresholds, and the continuous iteration of rules.

[0042] Furthermore, the intelligent identification of public opinion means through the public opinion warfare identification module specifically includes:

[0043] S301. Scenario grading mechanism;

[0044] Scenario grading includes the quantification of scenario features, and the quantification of scenario features includes the following indicators;

[0045] Propagation range indicator:

[0046] Platform coverage: Involves the number of platforms or the total number of monitored platforms;

[0047] Geographical distribution: Involves the number of regions or the total number of monitored regions;

[0048] User participation: The number of interactive users or the total number of exposed users;

[0049] Feature significance indicator:

[0050] Keyword matching degree: The number of hit keywords or the total number of keywords;

[0051] Topic relevance: The cosine similarity with the core topic;

[0052] Emotional polarity degree: The standard deviation of emotional values;

[0053] Influence indicator:

[0054] Propagation speed: The growth rate per unit time;

[0055] KOL participation: The weighted sum of the influence of participating KOLs;

[0056] Interaction intensity: The time-decaying weighted sum of comments and forwards;

[0057] Scenario judgment rules:

[0058] Regular scenario:

[0059] Platform coverage < 0.3;

[0060] Feature significance > 0.7;

[0061] Influence index < 0.4;

[0062] Handling strategy: Regular monitoring, response within 24 hours;

[0063] Complex scenarios:

[0064] 0.3 ≤ Platform coverage < 0.6;

[0065] 0.4 ≤ Feature significance ≤ 0.7;

[0066] 0.4 ≤ Influence index < 0.7;

[0067] Handling strategy: Key attention, response within 12 hours;

[0068] Emergency scenarios:

[0069] Platform coverage ≥ 0.6;

[0070] Feature significance < 0.4;

[0071] Influence index ≥ 0.7;

[0072] Handling strategy: Immediate disposal, response within 2 hours;

[0073] Dynamic adjustment mechanism:

[0074] Real-time update: Update the index value every 5 minutes;

[0075] Threshold optimization: Adjust the threshold once a week based on historical data;

[0076] Manual intervention: Support manual adjustment of the classification results;

[0077] Emergency response: Quick escalation mechanism in case of setting;

[0078] S302. Identification specification definition;

[0079] The identification specification includes:

[0080] Input specification, which includes basic event information, verified factual content, and public opinion content to be analyzed;

[0081] Analysis requirements, which include feature matching analysis, fact comparison and verification, impact degree assessment, and comprehensive judgment;

[0082] Output specification, which includes identification results, evidence explanation, and disposal suggestions;

[0083] The identification results include the type of public opinion warfare method and confidence level, the evidence explanation includes the deviation between features and facts, and the disposal suggestions include impact assessment and response plan;

[0084] Feature matching analysis includes:

[0085] Feature matching of public opinion warfare methods:

[0086] Matching with the type of fanning the flames, including calculating text deviation, evaluating emotional intensity, analyzing communication impact, and statistically analyzing keyword features;

[0087] Matching with the type of framing, including evaluating fact consistency, calculating target directivity, analyzing causal credibility, and calculating comprehensive scores;

[0088] Formal definition matching:

[0089] Basic feature comparison, including calculating text semantic similarity, calculating core element coverage rate, and determining dimension compliance;

[0090] Quantitative index matching, including index value calculation, threshold judgment, weight assignment, and comprehensive score calculation;

[0091] Factual comparison and verification include:

[0092] Basic fact verification, and the basic fact verification includes time dimension, space dimension, and subject dimension;

[0093] The time dimension includes:

[0094] Accuracy of event occurrence time, integrity of time series, temporal relationship of key nodes, and rationality of time span;

[0095] The space dimension includes:

[0096] Accuracy of location information, rationality of spatial range, verification of geographical relevance, and authenticity of scene description;

[0097] The subject dimension includes:

[0098] Verification of person identity, verification of organizational structure, sorting out relationship network, and confirmation of role positioning;

[0099] Content consistency verification, and the content consistency verification includes fact element comparison and logical relationship verification;

[0100] The fact element comparison includes core information matching degree, accuracy of detailed description, reliability of data citation, and evaluation of source credibility;

[0101] The logical relationship verification includes rationality of causal relationship, integrity of reasoning process, rigor of argument logic, and evaluation of conclusion reliability;

[0102] Impact degree evaluation includes: evaluation of communication dimension, evaluation of public opinion impact, and evaluation of social impact;

[0103] The evaluation of communication dimension includes: platform influence and user participation;

[0104] The platform influence includes mainstream media coverage, social media activity, self-media dissemination power, and calculation of the comprehensive influence score;

[0105] The user engagement includes evaluation of the interaction volume, analysis of the user portrait, analysis of the participation depth, and evaluation of the KOL influence;

[0106] The public opinion influence evaluation includes: sentiment tendency analysis, and issue development evaluation;

[0107] The sentiment tendency analysis includes sentiment polarity distribution, sentiment intensity change, group mood evolution, and polarization degree evaluation;

[0108] The issue development evaluation includes topic heat trend, issue diffusion range, related topic linkage, and development trend prediction;

[0109] The social influence evaluation includes: cognitive influence, and behavioral influence;

[0110] The cognitive influence includes the degree of public cognitive bias, the change trend of group attitude, the degree of value conflict, and the influence degree of social consensus;

[0111] The behavioral influence includes online behavior change, offline action conversion, group behavior trend, and social governance influence;

[0112] The comprehensive judgment includes: multi-dimensional scoring, and risk warning;

[0113] The multi-dimensional scoring includes calculation of the feature matching score, fact verification score, influence evaluation score, calculation of the comprehensive score and grade division;

[0114] The risk warning includes risk level determination, development trend prediction, generation of intervention suggestions, and effect evaluation mechanism;

[0115] S303. Result processing mechanism;

[0116] The said result processing mechanism includes confidence evaluation and processing strategy;

[0117] The said confidence evaluation includes evaluation of the feature matching degree, evaluation of the evidence integrity, evaluation of the influence degree, and calculation of the comprehensive confidence;

[0118] The said evaluation of the feature matching degree includes semantic feature matching degree, sentiment feature matching degree, and fact feature matching degree;

[0119] The said semantic feature matching degree includes cosine similarity with the standard pattern, key concept coverage rate, and semantic coherence score;

[0120] The said sentiment feature matching degree includes sentiment polarity consistency and sentiment intensity similarity;

[0121] The fact feature matching degree includes entity alignment rate and event element matching degree;

[0122] The evidence integrity evaluation includes evidence chain integrity and evidence quality;

[0123] The evidence chain integrity includes key node coverage rate, logical link integrity, and temporal coherence;

[0124] The evidence quality includes source reliability score, evidence freshness, and degree of mutual verification of evidence;

[0125] The impact degree evaluation includes communication influence and timeliness impact;

[0126] The communication influence includes platform weight score, user influence, and communication scope;

[0127] The timeliness impact includes real-time popularity and development trend;

[0128] The comprehensive confidence calculation includes weighted average calculation, and the calculation method is: Confidence = 0.4 * feature matching degree + 0.35 * evidence integrity + 0.25 * impact degree. Confidence represents the confidence level, and the confidence level classification is as follows: Confidence greater than or equal to 0.8 is high confidence level, Confidence between 0.6 and 0.8 is medium confidence level, and Confidence less than 0.6 is low confidence level;

[0129] The specific processing strategy includes: directly adopting for high confidence level, conducting manual review for medium confidence level, and re-analyzing for low confidence level;

[0130] S304. Optimization feedback;

[0131] The optimization feedback includes result accumulation and continuous optimization;

[0132] The result accumulation includes typical case warehousing, feature library update, and summary of handling experience;

[0133] The continuous optimization includes adjustment of grading standards, improvement of identification specifications, and optimization of processing strategies.

[0134] Furthermore, the confirmation and correction of the recognition result through the human-machine collaborative verification module specifically includes:

[0135] Audit process. The audit process includes a three-level audit mechanism of preliminary review, re-review, and final review. The average audit time for a single piece of content does not exceed 2 minutes. It supports real-time update of the prompt word template, regularly updates the knowledge base every week, and supports an emergency update mechanism. The preliminary review focuses on basic feature matching, the re-review emphasizes the integrity of the evidence chain, and the final review ensures the reliability of the judgment result;

[0136] Feedback mechanism, the feedback mechanism includes real-time feedback of review results, continuous optimization and update of the model, dynamic expansion of the knowledge base, and continuous improvement of system performance.

[0137] The beneficial effects of the present invention are as follows:

[0138] Through the formal definition of public opinion warfare methods and the quantitative index system, the present invention greatly improves the recognition accuracy. With the help of the multi-dimensional fact comparison mechanism, for the public opinion content that is reliably recognized as inconsistent with the real situation, the error rate is reduced. Relying on the human-machine collaborative verification mechanism, the reliability of the recognition result is ensured, and the comprehensive accuracy is improved.

[0139] Through the standardized public opinion warfare method recognition process, the present invention reduces the analysis time of a single piece of content. Through the optimized resource scheduling mechanism, it supports concurrent requests of more than multiple per second. The reuse mechanism of the scenario template improves the analysis efficiency.

[0140] Through the complete evidence chain construction mechanism, the present invention improves the traceability of the analysis process. Through the standardized result display scheme, it is convenient for professionals to understand and verify. The visual analysis process display improves the credibility of the system.

[0141] Through the formal knowledge base construction method, the present invention supports rapid adaptation to new public opinion warfare methods. The template dynamic optimization mechanism realizes flexible adjustment of strategies. The hierarchical system architecture design supports horizontal expansion and rapid capacity expansion.

[0142] The present invention reduces the manual review cost through the standardized process, reduces the system operation and maintenance cost through the optimized resource scheduling, and improves the public opinion handling efficiency through the efficient recognition mechanism. Brief Description of the Drawings

[0143] Figure 1 is the overall architecture diagram of the public opinion means intelligent recognition system provided by the embodiment of the present invention;

[0144] Figure 2 is the public opinion warfare method recognition flow chart provided by the embodiment of the present invention;

[0145] Figure 3 is the scenario classification recognition flow chart provided by the embodiment of the present invention;

[0146] Figure 4 is the prompt word template structure diagram provided by the embodiment of the present invention;

[0147] Figure 5 is the human-machine collaborative verification flow chart provided by the embodiment of the present invention. Detailed Embodiments

[0148] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention.

[0149] The present invention provides an intelligent recognition system for public opinion means based on an artificial intelligence large model, as Figure 1 shown, including:

[0150] A data layer, which includes a public opinion data storage module, a knowledge base storage module, and a result cache module. The public opinion data storage module stores the original public opinion data using a distributed database. The knowledge base storage module stores the public opinion warfare knowledge system using a graph database. The result cache module stores the analysis results using a cache.

[0151] A service layer, which includes a data collection and preprocessing module, an event element extraction module, a fact verification module, an opinion warfare recognition module, and a human-machine collaborative verification module. The data collection and preprocessing module is used to obtain, clean, and standardize the data of public opinion events. The event element extraction module is used to identify and extract the core element information of public opinion events. The fact verification module is used to confirm the true development context of the event through manual review. The opinion warfare recognition module is used to intelligently identify public opinion means based on a large model. The human-machine collaborative verification module is used to support professionals in confirming and correcting the recognition results.

[0152] An interface layer, which includes a standard service interface, a real-time push interface, and an asynchronous task interface. The standard service interface is used to provide standardized service calls. The real-time push interface is used to support real-time data push. The asynchronous task interface is used to process asynchronous tasks.

[0153] The specific implementation process of the present invention includes the following steps:

[0154] Step 1, initialization of public opinion events;

[0155] Determine the basic theme and scope of the event to be analyzed;

[0156] Extract the key elements of the event, including but not limited to time, place, person, event process, etc.;

[0157] Determine the true situation and development context of the event through system extraction and manual review.

[0158] Step 2, construct a public opinion warfare knowledge base;

[0159] S201, formal definition of public opinion warfare;

[0160] For the type of fanning the flames, the formal definition is as follows:

[0161] Basic features: Exaggeration, embellishment, or over-interpretation during the spread of public opinion events;

[0162] Core elements: Secondary spread, emotional embellishment, and influence expansion;

[0163] Judgment dimensions: Temporal characteristics, content characteristics, and communication characteristics;

[0164] Quantitative index system:

[0165] Text deviation degree: Similarity threshold with the original facts;

[0166] Emotional intensity: Proportion of emotion words and polarity value threshold;

[0167] Communication influence: Topic growth rate and communication scope threshold;

[0168] Keyword characteristics: Threshold of the frequency of inciting words;

[0169] Comprehensive score: Multi-dimensional weighted calculation method;

[0170] The type of framing and false accusation is formally defined as follows:

[0171] Basic features: The act of fabricating or distorting facts and attributing them to a specific object;

[0172] Core elements: False information, causal distortion, and target orientation;

[0173] Judgment dimensions: Fact deviation, target orientation, and causal relationship;

[0174] Quantitative index system:

[0175] Fact consistency: Matching degree threshold with the verified facts;

[0176] Target orientation: Threshold of the frequency of mention of a specific object;

[0177] Causal credibility: Semantic relationship analysis threshold;

[0178] Comprehensive score: Multi-dimensional weighted calculation method;

[0179] S202. Build a knowledge base for public opinion warfare methods;

[0180] Build a basic knowledge system, including the classification system of public opinion warfare method types, formal definition specification standards, quantitative index calculation methods, and evidence chain construction rules;

[0181] Case library construction, including typical case annotation specifications, case feature extraction methods, case classification indexing mechanisms, and similar case matching rules;

[0182] Dynamic update mechanism, including extraction of new combat method features, dynamic adjustment of the indicator system, adaptive optimization of thresholds, and continuous iteration of rules.

[0183] Step 3, intelligent identification of public opinion combat methods;

[0184] Intelligent identification of public opinion combat methods is as Figure 2 shown and includes the following steps:

[0185] S301, scenario classification mechanism;

[0186] Scenario classification includes scenario feature quantification, scenario judgment rules, and dynamic adjustment mechanism;

[0187] Scenario feature quantification includes the following indicators;

[0188] Propagation range indicator:

[0189] Platform coverage: related to the number of platforms or the total number of monitored platforms;

[0190] Geographical distribution: related to the number of regions or the total number of monitored regions;

[0191] User participation: the number of interactive users or the total number of exposed users;

[0192] Feature significance indicator:

[0193] Keyword matching degree: the number of hit keywords or the total number of keywords;

[0194] Topic relevance: the cosine similarity with the core topic;

[0195] Emotional polarity degree: the standard deviation of the emotional value;

[0196] Influence indicator:

[0197] Propagation speed: the growth rate per unit time;

[0198] KOL participation: the weighted sum of the influence of participating KOLs;

[0199] Interaction intensity: the time-decayed weighted sum of comments and forwards;

[0200] Scenario judgment rules, as Figure 3 shown:

[0201] Regular scenario (low risk):

[0202] Platform coverage < 0.3;

[0203] Feature significance > 0.7;

[0204] Influence indicator < 0.4;

[0205] Processing strategy: Regular monitoring, response within 24 hours;

[0206] Complex scenarios (medium risk):

[0207] 0.3 ≤ Platform coverage < 0.6;

[0208] 0.4 ≤ Feature significance ≤ 0.7;

[0209] 0.4 ≤ Influence index < 0.7;

[0210] Processing strategy: Key attention, response within 12 hours;

[0211] Emergency scenarios (high risk):

[0212] Platform coverage ≥ 0.6;

[0213] Feature significance < 0.4;

[0214] Influence index ≥ 0.7;

[0215] Processing strategy: Immediate disposal, response within 2 hours;

[0216] Dynamic adjustment mechanism:

[0217] Real-time update: Update the index value every 5 minutes;

[0218] Threshold optimization: Adjust the threshold once a week based on historical data;

[0219] Manual intervention: Support manual adjustment of the classification results;

[0220] Emergency response: Quick escalation mechanism under certain settings;

[0221] S302. Identification specification definition;

[0222] The identification specification includes:

[0223] Input specification, which includes basic event information, verified factual content, and public opinion content to be analyzed;

[0224] Analysis requirements, which include feature matching analysis, factual comparison and verification, impact degree assessment, and comprehensive judgment;

[0225] Output specification, which includes identification results, evidence description, and disposal suggestions;

[0226] The identification results include the type of public opinion warfare method and confidence level, the evidence description includes the deviation between features and facts, and the disposal suggestions include impact assessment and response plans;

[0227] The feature matching analysis includes:

[0228] Feature matching of public opinion warfare methods:

[0229] Match with the type of fanning the flames, including calculating the text deviation degree (similarity with the original fact), evaluating the emotional intensity (proportion of emotion words and polarity value), analyzing the communication impact (topic growth rate and scope), and counting the keyword features (frequency of inciting words);

[0230] Match with the type of framing, including evaluating the fact consistency (matching degree with the verified fact), calculating the target directivity (mention frequency of specific objects), analyzing the causal credibility (semantic relationship analysis), and calculating the comprehensive score (multi-dimensional weighting);

[0231] Formal definition matching:

[0232] Basic feature comparison, including calculating the text semantic similarity (using the BERT model), calculating the core element coverage rate (effective match when greater than or equal to 85%), and determining the dimension compliance degree (effective match when greater than or equal to 80%);

[0233] Quantitative index matching, including index value calculation (according to the specific public opinion warfare method type), threshold judgment (comparison with the preset standard), weight assignment (importance of different dimensions), and comprehensive score calculation (weighted average method);

[0234] Fact comparison and verification include:

[0235] Basic fact verification, and the basic fact verification includes time dimension, space dimension, and subject dimension;

[0236] The time dimension includes:

[0237] Accuracy of the event occurrence time, integrity of the time series, timing relationship of key nodes, and rationality of the time span;

[0238] The space dimension includes:

[0239] Accuracy of the location information, rationality of the space range, verification of geographical relevance, and authenticity of the scene description;

[0240] The subject dimension includes:

[0241] Verification of the identity of the person, verification of the organization, sorting out of the relationship network, and confirmation of the role positioning;

[0242] Content consistency verification, and the content consistency verification includes fact element comparison and logical relationship verification;

[0243] The fact element comparison includes the core information matching degree (greater than or equal to 90%), the accuracy of the detailed description (greater than or equal to 85%), the reliability of the data citation (greater than or equal to 95%), and the evaluation of the source credibility (greater than or equal to 80%);

[0244] Logical relationship verification includes the rationality of causal relationships, the integrity of reasoning processes, the rigor of argumentation logic, and the assessment of conclusion reliability;

[0245] Impact degree assessment includes: dissemination dimension assessment, public opinion impact assessment, and social impact assessment;

[0246] Dissemination dimension assessment includes: platform influence and user participation;

[0247] Platform influence includes mainstream media coverage (weight: 0.4), social media activity (weight: 0.3), self-media dissemination power (weight: 0.3), and calculation of comprehensive influence scores;

[0248] User participation includes assessment of interaction volume, user portrait analysis, analysis of participation depth, and assessment of KOL influence;

[0249] Public opinion impact assessment includes: sentiment tendency analysis and issue development assessment;

[0250] Sentiment tendency analysis includes sentiment polarity distribution (positive, negative, neutral), change in sentiment intensity (temporal trend), evolution of group sentiment (dissemination path), and assessment of polarization degree (viewpoint distribution);

[0251] Issue development assessment includes topic heat trend (growth rate, persistence), scope of issue diffusion (region, population), linkage of related topics (topic cluster analysis), and prediction of development trend (trend model);

[0252] Social impact assessment includes: cognitive impact and behavioral impact;

[0253] Cognitive impact includes the degree of public cognitive bias, trend of group attitude change, degree of value conflict, and impact on social consensus;

[0254] Behavioral impact includes changes in online behavior (interaction method, participation degree), conversion of offline actions (field activities, social mobilization), trend of group behavior (herd effect, polarization phenomenon), and impact on social governance (policy response, institutional adjustment);

[0255] Comprehensive judgment includes: multi-dimensional scoring and risk warning;

[0256] Multi-dimensional scoring includes calculation of feature matching scores (weight: 0.4), fact verification scores (weight: 0.35), impact assessment scores (weight: 0.25), calculation of comprehensive scores and grade division;

[0257] Risk warning includes determination of risk levels (high, medium, low), prediction of development trends (24h, 3 days, 7 days), generation of intervention suggestions (based on historical cases), and effect assessment mechanism (feedback after intervention);

[0258] S303, Result Processing Mechanism;

[0259] The result processing mechanism includes confidence evaluation and processing strategies;

[0260] Confidence evaluation includes feature matching degree evaluation (weight: 0.4), evidence integrity evaluation (weight: 0.35), impact degree evaluation (weight: 0.25), and comprehensive confidence calculation;

[0261] Feature matching degree evaluation includes semantic feature matching degree, sentiment feature matching degree, and fact feature matching degree;

[0262] Semantic feature matching degree includes cosine similarity with the standard pattern (threshold > 0.8), key concept coverage rate (threshold > 0.7), and semantic coherence score (threshold > 0.75);

[0263] Sentiment feature matching degree includes sentiment polarity consistency (threshold > 0.8) and sentiment intensity similarity (threshold > 0.7);

[0264] Fact feature matching degree includes entity alignment rate (threshold > 0.9) and event element matching degree (threshold > 0.85);

[0265] Evidence integrity evaluation includes evidence chain integrity and evidence quality;

[0266] Evidence chain integrity includes key node coverage rate (requirement > 90%), logical link integrity (requirement > 85%), and temporal coherence (requirement > 95%);

[0267] Evidence quality includes source reliability score (full score 100, requirement > 80), evidence freshness (within 7 days: 1.0, within 14 days: 0.8, within 30 days: 0.6), and evidence mutual verification degree (requirement > 3 independent sources);

[0268] Impact degree evaluation includes communication influence and timeliness influence;

[0269] Communication influence includes platform weight score (mainstream media: 1.0, social media: 0.8, others: 0.6), user influence (KOL: 1.0, ordinary users: 0.6), and communication scope (national: 1.0, regional: 0.7, local: 0.4);

[0270] Timeliness influence includes real-time popularity (within 24 hours: 1.0, within 3 days: 0.8, within 7 days: 0.6) and development trend (rising: 1.0, stable: 0.7, falling: 0.4);

[0271] The comprehensive confidence calculation includes weighted average calculation, and the calculation method is: Confidence = 0.4 * Feature matching degree + 0.35 * Evidence integrity + 0.25 * Impact degree. Confidence represents the confidence level, and the confidence level classification is as follows: Confidence greater than or equal to 0.8 is high confidence, Confidence between 0.6 and 0.8 is medium confidence, and Confidence less than 0.6 is low confidence;

[0272] The specific processing strategy includes: directly adopting high-confidence results, conducting manual review for medium-confidence results, and re-analyzing low-confidence results;

[0273] S304. Optimization feedback;

[0274] Optimization feedback includes result accumulation and continuous optimization;

[0275] Result accumulation includes storing typical cases in the database, updating the feature library, and summarizing disposal experience; Continuous optimization includes adjusting the grading standard, improving the recognition specification, and optimizing the processing strategy.

[0276] Step 4. Human-machine collaborative verification;

[0277] Human-machine collaborative verification includes:

[0278] (1) Review process:

[0279] A three-level review mechanism of preliminary review -> re-review -> final review;

[0280] The average review time for a single piece of content does not exceed 2 minutes;

[0281] Support real-time update of the prompt word template;

[0282] Regularly update the knowledge base every week, and support the emergency update mechanism.

[0283] (2) Review criteria:

[0284] Preliminary review: Focus on basic feature matching;

[0285] Re-review: Focus on the integrity of the evidence chain;

[0286] Final review: Ensure the reliability of the judgment result.

[0287] (3) Feedback mechanism:

[0288] The review result is feedback in real time;

[0289] The model is continuously optimized and updated;

[0290] The knowledge base is dynamically expanded;

[0291] The system performance is continuously improved.

[0292] The key technical points of the present invention include:

[0293] I. Prompt Engineering:

[0294] The prompt templates are as Figure 4 shown:

[0295] (1) Formal description framework:

[0296] Structured representation of the characteristics of public opinion warfare methods;

[0297] Quantitative index system for judgment criteria;

[0298] Specification definition of evidence requirements;

[0299] Standardized description of the analysis process;

[0300] (2) Scenario-based template design:

[0301] Special templates for different types of public opinion warfare methods;

[0302] Template verification mechanism based on cases;

[0303] Template performance evaluation system;

[0304] Dynamic optimization feedback mechanism;

[0305] (3) Specification for constructing evidence chain:

[0306] Key feature extraction method;

[0307] Evidence integrity requirements;

[0308] Credibility evaluation criteria;

[0309] Result traceability mechanism.

[0310] II. Fact comparison mechanism:

[0311] (1) Text similarity calculation:

[0312] a. Semantic vector model calculation:

[0313] Text encoding:

[0314] Encode the text into a 768-dimensional vector using the BERT model;

[0315] Adopt hierarchical pooling to obtain sentence representation;

[0316] Use the attention mechanism to weight key information;

[0317] Output normalization processing;

[0318] Similarity calculation:

[0319] Cosine similarity: cos(θ) = (A·B) / (||A||·||B||);

[0320] Euclidean distance: d = sqrt(Σ(ai - bi)2);

[0321] Manhattan distance: d = Σ|ai - bi|;

[0322] Weighted combined score: Score = w1*cos + w2*(1 - deuc) + w3*(1 - dman); b. Edit distance algorithm:

[0323] Levenshtein distance calculation:

[0324] Cost of insertion operation: 1.0;

[0325] Cost of deletion operation: 1.0;

[0326] Cost of substitution operation: 2.0;

[0327] Normalization: score = 1 - dist / max(len1, len2);

[0328] Longest common subsequence (LCS):

[0329] Calculating the length of LCS using dynamic programming;

[0330] Backtracking to construct the specific sequence;

[0331] Similarity score: score = LCS_length / max(len1, len2);

[0332] c. Key information extraction:

[0333] Named entity recognition (NER):

[0334] Identifying entities such as time, place, person, organization, etc.;

[0335] Calculating the entity matching rate: match_rate = matched_entities / total_entities; Weighting the entity importance: weight = entity_type_weight*entity_frequency; Relationship extraction:

[0336] Extracting subject-predicate-object triples;

[0337] Calculating the relationship matching degree;

[0338] Constructing an event graph;

[0339] Calculate the similarity of the atlas;

[0340] d. Comprehensive similarity calculation:

[0341] Feature fusion:

[0342] Semantic similarity weight: 0.4;

[0343] Edit distance weight: 0.3;

[0344] Entity matching weight: 0.2;

[0345] Relationship matching weight: 0.1;

[0346] Threshold determination:

[0347] Highly similar: score ≥ 0.8;

[0348] Partially similar: 0.6 ≤ score < 0.8;

[0349] Lowly similar: score < 0.6;

[0350] (2) Difference detection algorithm:

[0351] a. Rule-based explicit difference detection;

[0352] Text structure differences:

[0353] Paragraph structure comparison: Calculate the changes in the number, length, and position of paragraphs;

[0354] Syntactic structure comparison: Analyze the changes in sentence components and grammatical relationships;

[0355] Punctuation mark comparison: Identify the changes in punctuation usage, position, and frequency;

[0356] Format style comparison: Detect the changes in font, typesetting, and layout;

[0357] Content element differences:

[0358] Keyword differences: Addition, deletion, and modification of core vocabulary;

[0359] Numerical information differences: Changes in numbers, dates, and amounts;

[0360] Entity information differences: Changes in names of people, places, and organizations;

[0361] Event description differences: Changes in the process and details of events;

[0362] b. Semantic-based implicit difference detection;

[0363] Semantic deviation analysis:

[0364] Semantic level: synonym replacement, near-synonym variation;

[0365] Phrase level: change in expression mode, change in tone;

[0366] Sentence level: change in contextual meaning, change in logical relationship;

[0367] Paragraph level: deviation of the theme focus, change in the perspective of discussion;

[0368] Differences in emotional tendency:

[0369] Change in emotional polarity: transition between positive / negative / neutral;

[0370] Change in emotional intensity: change in the use of degree words;

[0371] Differences in subjectivity: objective statement vs. subjective evaluation;

[0372] Differences in stance and attitude: change between support / opposition / neutral;

[0373] c. Multi-dimensional comparative analysis;

[0374] Comparisons in the time dimension include consistency of chronological relationships, changes in time spans, precision of time points, and logical coherence of time; comparisons in the location dimension include accuracy of geographical locations, changes in spatial ranges, consistency of geographical relationships, and completeness of scene descriptions; comparisons in the character dimension include accuracy of character identities, changes in character relationships, consistency of behavior descriptions, and accuracy of speech citations; comparisons in the event dimension include changes in causal relationships, completeness of process descriptions, differences in result impacts, and changes in background information;

[0375] d. Difference assessment and handling;

[0376] Quantification of the degree of differences:

[0377] Explicit difference score: based on the results of rule detection;

[0378] Implicit difference score: based on the results of semantic analysis;

[0379] Comprehensive difference score: calculated by weighted average;

[0380] Difference classification:

[0381] Major differences: changes that affect the essence of the event;

[0382] General differences: changes in details or expressions;

[0383] Negligible differences: minor changes that do not affect understanding;

[0384] Handling strategies:

[0385] Major differences: require manual review;

[0386] General differences: Systematically marked;

[0387] Negligible differences: Passed directly;

[0388] III. Interpretability design techniques:

[0389] Interpretability design includes:

[0390] Visualization of the analysis process: Presenting the analysis process in the form of a decision tree; Construction of the evidence chain: Combining the highlighting of key information and the display of the reasoning path; a. Basic structure of the evidence chain:

[0391] Core fact layer: Basic facts of the original event;

[0392] Propagation layer: Information propagation process and changes;

[0393] Impact layer: Social impact and public opinion effect;

[0394] Verification layer: Fact verification and corroborating materials;

[0395] b. Extraction of key information:

[0396] Event elements:

[0397] Time information: Occurrence time, duration, key time nodes; Location information: Occurrence location, scope of influence, geographical correlation;

[0398] Person information: Parties involved, related parties, witnesses;

[0399] Organization information: Involved units, regulatory departments, media organizations;

[0400] Event process:

[0401] Cause: Triggering event, background conditions;

[0402] Process: Key actions, important nodes;

[0403] Result: Direct impact, indirect impact;

[0404] Development: Subsequent progress, derivative events;

[0405] Relevant information:

[0406] Related events: Historical cases, similar events;

[0407] Background information: Policies and regulations, industry standards;

[0408] Expert opinions: Authoritative interpretations, professional analyses;

[0409] Public reaction: Public opinion attitude, mass opinions;

[0410] c. Evidence grading and evaluation:

[0411] Evidence level:

[0412] Level 1 evidence: official documents, direct evidence (weight: 0.5); Level 2 evidence: reports by authoritative media, expert analysis (weight: 0.3); Level 3 evidence: circumstantial evidence, public feedback (weight: 0.2); Evidence quality evaluation:

[0413] Authenticity: reliability of the source, truthfulness of the content;

[0414] Timeliness: freshness of the information, timeliness of updates;

[0415] Relevance: degree of relevance to the core facts;

[0416] Completeness: degree of completeness of the information;

[0417] d. Example of evidence chain construction:

[0418] Taking a product quality incident as an example:

[0419] 1. Core fact layer:

[0420] Level 1 evidence:

[0421] Quality inspection report by the market supervision department;

[0422] Official statement of the enterprise;

[0423] On-site inspection record;

[0424] Evidence correlation:

[0425] Time: The problem was discovered on X month X, 2023;

[0426] Location: Specific production base and sales area;

[0427] Involved parties: Production enterprise, distributor;

[0428] Problem description: Specific quality problems and scope of influence;

[0429] 2. Dissemination layer:

[0430] Level 2 evidence:

[0431] Investigative reports by mainstream media;

[0432] Analysis articles by industry experts;

[0433] Investigative reports by consumer associations;

[0434] Information dissemination path:

[0435] Social media first release → media follow-up reports → official response;

[0436] Propagation timeline and nodes;

[0437] Evolution process of important viewpoints;

[0438] 3. Impact layer:

[0439] Tertiary evidence includes consumer complaints and feedback, social media discussions and comments, and changes in market sales data;

[0440] Impact assessment includes brand reputation impact, market share changes, and consumer confidence index;

[0441] 4. Verification layer:

[0442] Cross-verification includes multi-source information comparison, timeline consistency check, and mutual verification of key details;

[0443] Credibility assessment includes evidence integrity: 90%, source reliability: 85%, and logical consistency: 95%;

[0444] Result scoring mechanism: includes credibility scoring and evidence sufficiency assessment;

[0445] Audit feedback record: supports problem annotation and opinion feedback.

[0446] The above are only the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and environments, and can be changed within the scope of the concept described herein through the above teachings or the techniques or knowledge in related fields. Any changes and variations made by those skilled in the art without departing from the spirit and scope of the present invention shall fall within the protection scope of the appended claims of the present invention.

Claims

1. The intelligent recognition system of public opinion means based on the artificial intelligence large model is characterized by: It includes a data layer and a service layer. The data layer includes a public opinion data storage module, a knowledge base storage module and a result cache module. The public opinion data storage module uses a distributed database to store original public opinion data. The knowledge base storage module uses a graph database to store the public opinion warfare knowledge system. The result cache module uses a high-speed cache to store analysis results. The service layer includes a data collection and preprocessing module, an event element extraction module, a fact verification module, a public opinion tactics identification module and a human-machine collaborative verification module. The data collection and preprocessing module is used to obtain public opinion event data, and clean and standardize it. After the data is standardized, the event element extraction module is used to identify and extract the core element information of the public opinion event, and then the fact verification module is used to review and confirm the true development context of the event, and then the public opinion tactics identification module is used to intelligently identify public opinion means, and finally the human-machine collaborative verification module is used to confirm and correct the identification results.

2. The public opinion means intelligent identification system based on artificial intelligence large model according to claim 1 is characterized in that: The system also includes an interface layer, which includes a standard service interface, a real-time push interface and an asynchronous task interface. The standard service interface is used to provide standardized service calls, the real-time push interface is used to support real-time data push, and the asynchronous task interface is used to process asynchronous tasks.

3. The public opinion means intelligent identification system based on artificial intelligence large model according to claim 1 is characterized in that: The event element extraction module is used to identify and extract the core element information of public opinion events, including: The event element extraction module is used to determine the basic theme and scope of the public opinion event to be analyzed, and to extract the key elements of the public opinion event, including time, place, people, and the course of events.

4. The public opinion means intelligent identification system based on artificial intelligence large model according to claim 1 is characterized in that: The public opinion tactics identification module is also used to build a public opinion tactics knowledge base. The construction of the public opinion tactics knowledge base specifically includes: S201. Conduct a formal definition of public opinion warfare tactics; The type of fueling is formally defined as follows: Basic characteristics: exaggeration, rendering or over-interpretation during the dissemination of public opinion events; Core elements: secondary dissemination, emotional rendering, and influence expansion; Determination dimensions: time series characteristics, content characteristics, and communication characteristics; Quantitative indicator system: Text deviation: similarity threshold with the original facts; Sentiment intensity: emotional word proportion and polarity value threshold; Dissemination impact: topic growth rate and dissemination range threshold; Keyword features: frequency threshold of inflammatory words; Comprehensive score: multi-dimensional weighted calculation method; The types of frame-up are formally defined as follows: Basic characteristics: the act of fabricating or distorting facts to attribute blame to a specific person; Core elements: false information, causal distortion, and target orientation; Judgment dimensions: factual deviation, target orientation, and causal relationship; Quantitative indicator system: Factual consistency: the matching threshold with the verified facts; Target-directedness: frequency threshold of specific object mentions; Causal credibility: threshold for semantic relationship analysis; Comprehensive score: multi-dimensional weighted calculation method; S202. Build a knowledge base of public opinion tactics; Construct a basic knowledge system, including a classification system for public opinion warfare types, formal definition standards, quantitative indicator calculation methods, and evidence chain construction rules; Case database construction, including typical case annotation specifications, case feature extraction methods, case classification indexing mechanism, and similar case matching rules; Dynamic update mechanism, including extraction of new tactics features, dynamic adjustment of indicator system, adaptive optimization of thresholds, and continuous iteration of rules.

5. The public opinion means intelligent identification system based on artificial intelligence large model according to claim 1 is characterized in that: Intelligent identification of public opinion tactics through the public opinion tactics identification module includes: S301, scene classification mechanism; Scene classification includes scene feature quantification, which includes the following indicators: Reach indicators: Platform coverage: the number of platforms involved or the total number of monitored platforms; Geographical distribution: the number of regions involved or the total number of monitored regions; User engagement: the number of interactive users or the total number of exposed users; Feature significance index: Keyword matching degree: the number of hit keywords or the total number of keywords; Topic relevance: cosine similarity with the core topic; Sentiment polarity: standard deviation of sentiment values; Impact indicators: Propagation speed: the growth rate per unit time; KOL participation: the weighted sum of the influence of participating KOLs; Interaction intensity: the time-decayed weighted sum of comments and reposts; Scene judgment rules: Common scenarios: Platform coverage < 0.3; Feature significance > 0.7; Influence index <0.4; Processing strategy: routine monitoring, response within 24 hours; Complex scenarios: 0.3≤Platform coverage<0.6; 0.4≤Feature significance≤0.7; 0.4≤Influence Index<0.7; Processing strategy: Pay close attention and respond within 12 hours; Emergency scenarios: Platform coverage ≥ 0.6; Feature significance < 0.4; Influence index ≥ 0.7; Processing strategy: Immediate treatment, response within 2 hours; Dynamic adjustment mechanism: Real-time update: indicator values ​​are updated every 5 minutes; Threshold optimization: adjust the threshold once a week based on historical data; Manual intervention: support manual adjustment of grading results; Emergency response: Rapid escalation mechanism under set circumstances; S302, identification specification definition; Identification specifications include: Input specifications, including basic information of the event, verified facts, and public opinion content to be analyzed; Analysis requirements, including feature matching analysis, fact comparison verification, impact assessment, and comprehensive research and judgment; Output specifications, including identification results, evidence description, and disposal suggestions; The identification results include the type and confidence level of the public opinion tactics, the evidence description includes characteristics and factual deviations, and the disposal recommendations include impact assessment and response plans; Feature matching analysis includes: Public Opinion Tactics Feature Matching: Matching with the type of promotion, including calculating text deviation, evaluating sentiment intensity, analyzing communication impact, and counting keyword features; Matching with the frame-up type, including assessing factual consistency, calculating target orientation, analyzing causal credibility, and calculating comprehensive scores; Formal definition matching: Basic feature comparison, including calculating text semantic similarity, calculating core element coverage, and determining dimension conformity; Quantitative indicator matching, including indicator value calculation, threshold judgment, weight allocation, and comprehensive score calculation; Fact comparison verification includes: Basic fact verification, the basic fact verification includes time dimension, space dimension, and subject dimension; The time dimension includes: The accuracy of the time of event occurrence, the completeness of the time series, the timing relationship of key nodes, and the rationality of the time span; The spatial dimensions include: The accuracy of location information, rationality of spatial scope, verification of geographical relevance, and authenticity of scene description; The subject dimensions include: Verify identity of people, verify organizational structure, sort out relationship networks, and confirm role positioning; Content consistency verification, including factual element comparison and logical relationship verification; The factual element comparison includes the matching degree of core information, accuracy of detailed description, reliability of data citation, and source credibility assessment; The verification of logical relationships includes the rationality of causal relationships, the integrity of the reasoning process, the rigor of argumentation logic, and the reliability assessment of conclusions; Impact assessment includes: communication dimension assessment, public opinion impact assessment, and social impact assessment; The communication dimension evaluation includes: platform influence and user engagement; The platform influence includes mainstream media coverage, social media activity, self-media communication power, and comprehensive influence score calculation; User engagement includes interaction level assessment, user portrait analysis, engagement depth analysis, and KOL influence assessment; Public opinion impact assessment includes: sentiment tendency analysis, topic development assessment; Emotional tendency analysis includes emotional polarity distribution, emotional intensity changes, group emotional evolution, and polarization degree assessment; Topic development assessment includes topic popularity trends, topic diffusion range, related topic linkage, and development trend prediction; Social impact assessment includes: cognitive impact, behavioral impact; Cognitive influence includes the degree of public cognitive bias, the trend of group attitude change, the degree of value conflict, and the influence of social consensus; Behavioral impacts include online behavior changes, offline action conversions, group behavior trends, and social governance impacts; Comprehensive assessment includes: multi-dimensional scoring and risk warning; Multi-dimensional scoring includes calculation of feature matching score, fact verification score, impact assessment score, comprehensive score calculation and grading; Risk early warning includes risk level determination, development trend prediction, intervention suggestion generation, and effect evaluation mechanism; S303, result processing mechanism; The result processing mechanism includes confidence assessment and processing strategy; The confidence assessment includes feature matching degree assessment, evidence integrity assessment, impact degree assessment, and comprehensive confidence calculation; The feature matching degree evaluation includes semantic feature matching degree, emotional feature matching degree, and factual feature matching degree; The semantic feature matching degree includes cosine similarity with the standard pattern, key concept coverage, and semantic coherence score; The emotional feature matching degree includes emotional polarity consistency and emotional intensity similarity; The fact feature matching degree includes entity alignment rate and event element matching degree; The evidence integrity assessment includes the integrity of the chain of evidence and the quality of evidence; The integrity of the chain of evidence includes key node coverage, logical link integrity, and temporal coherence; The quality of evidence includes source reliability score, freshness of evidence, and degree of mutual confirmation of evidence; The impact assessment includes communication influence and timeliness impact; The communication influence includes platform weight score, user influence, and communication scope; The timeliness impact includes real-time popularity and development trends; The comprehensive confidence calculation includes weighted average calculation, which is calculated as follows: Confidence = 0.4*feature matching degree + 0.35*evidence integrity + 0.25*influence degree, where Confidence represents confidence, and the confidence levels are as follows: Confidence greater than or equal to 0.8 is high confidence, Confidence between 0.6 and 0.8 is medium confidence, and Confidence less than 0.6 is low confidence; The processing strategy specifically includes: directly accepting the high confidence level, manually reviewing the medium confidence level, and re-analyzing the low confidence level; S304, optimization feedback; The optimization feedback includes result accumulation and continuous optimization; The accumulation of results includes storage of typical cases, updating of feature database, and summary of treatment experience; The continuous optimization includes adjustment of classification standards, improvement of identification specifications, and optimization of processing strategies.

6. The public opinion means intelligent identification system based on artificial intelligence large model according to claim 1 is characterized in that: Confirming and correcting the recognition results through the human-machine collaborative verification module specifically includes: The review process includes a three-level review mechanism of preliminary review, re-review and final review. The average review time for a single piece of content does not exceed 2 minutes. It supports real-time update of prompt word templates, regular weekly updates of the knowledge base, and supports emergency update mechanisms. The preliminary review focuses on basic feature matching, the re-review focuses on the integrity of the evidence chain, and the final review ensures the reliability of the judgment results; Feedback mechanism, which includes real-time feedback of audit results, continuous optimization and updating of models, dynamic expansion of knowledge base, and continuous improvement of system performance.

Images