Public opinion intelligent deduction and guidance method and system based on evolution modeling

Abstract

The invention discloses a public opinion intelligent deduction and guidance method and system based on evolution modeling, and belongs to the technical field of artificial intelligence. The method comprises the following steps: establishing a negative emotion dynamic evolution model according to a negative emotion text; and obtaining a critical time node of an evolution stage by using the negative emotion dynamic evolution model. According to the method, through the dynamic evolution model, accurate recognition of negative emotional evolution key time nodes is achieved, collaborative simulation deduction of strategy'time-degree-efficiency 'parameters is guided, and through a classification scene strategy generation and parameter dynamic optimization mechanism, the adaptation capability of the system to multiple types of emergencies is improved.

Description

Technical Field

[0001] The present invention relates to a method and system for intelligent deduction and guidance of public opinion based on evolutionary modeling, and belongs to the field of artificial intelligence technology. Background Art

[0002] Existing technologies often rely on empirical rules (such as the "golden four hours") to identify the timing of negative sentiment guidance, lacking precise modeling support based on dynamic public opinion data. For example, traditional methods fail to consider the quantitative delineation of negative sentiment evolutionary stages, making it impossible to scientifically predict the optimal intervention window (such as the critical point between pre- and post-guidance). This can lead to missed critical response opportunities or premature intervention, triggering a public opinion backlash.

[0003] Existing technologies often rely on static models (such as single-shot emotion recognition or single propagation models), making it difficult to track the evolution of negative sentiment in real time. Especially in the context of public opinion big data, the dynamic nature of sentiment demands continuous iteration of guidance strategies. However, traditional methods, due to issues such as data update lags and rigid model parameters, cannot achieve coordinated dynamic optimization of "time, scale, and effectiveness."

[0004] Existing technologies for evaluating the effectiveness of guidance are often based on post-event public opinion data statistics (such as changes in sentiment polarity ratios), lacking models for predicting guidance effectiveness based on real-time big data. This makes it impossible to quantitatively predict the direct impact of different "time-efficacy" combinations on the suppression of negative emotions before guidance measures are implemented, leading to blind spots in strategy formulation.

[0005] The existing technology is not based on the Logistic model, cannot dynamically adjust the guidance coefficient α, and is predicted to generate secondary risks. Summary of the Invention

[0006] In order to overcome the shortcomings of existing technologies, the purpose of the present invention is to provide a method and system for intelligent deduction and guidance of public opinion based on evolutionary modeling. Through a dynamic evolutionary model, accurate identification of key time nodes in the evolution of negative emotions can be achieved, and the system's adaptability to multiple types of emergencies can be improved through classification scenario strategy generation and parameter dynamic optimization mechanism.

[0007] To achieve the above-mentioned purpose, the present invention provides a method for intelligent deduction and guidance of public opinion based on evolutionary modeling, which comprises the following steps: S1: Establish a negative sentiment dynamic evolution model based on negative sentiment texts; S2: Use the negative emotion dynamic evolution model to obtain the critical time nodes of the evolution stage.

[0008] To achieve the aforementioned purpose of the invention, the present invention also provides a system comprising a storage medium and one or more processors, wherein the storage medium is used to store code for compiling the above-mentioned method of intelligent deduction and guidance of public opinion based on evolutionary modeling into a computer program using a computer language, and the computer program can be called and executed by one or more processors.

[0009] Compared to existing technologies, the proposed method and system for intelligently inferring and guiding public opinion based on evolutionary modeling accurately identifies key time points in the evolution of negative sentiment through a dynamic evolutionary model. Furthermore, the system's adaptability to multiple types of emergencies is enhanced through the generation of classified scenario strategies and dynamic parameter optimization. This method can be used in intelligent information governance scenarios for government public administration, online public opinion monitoring, emergency response, and social media platforms, providing real-time decision support for guiding negative sentiment. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flow chart of the method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention.

[0011] Figure 2 This is the negative emotion dynamic evolution model provided by the present invention. DETAILED DESCRIPTION

[0012] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0013] Figure 1 This is a flow chart of the method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention, such as Figure 1 As shown, the method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention includes the following steps: S1: Establish a negative sentiment dynamic evolution model based on negative sentiment texts; S2: Use the negative emotion dynamic evolution model to obtain the critical time nodes of the evolution stage.

[0014] In the present invention, establishing a negative emotion dynamic evolution model based on negative emotion text includes: S1-1: Building a regression model for: , Where, is the growth rate, K is the information limit; It is the amount of negative emotional information at the initial moment of negative emotion; S1-2: Regression Model Train to get the best and .

[0015] For regression models Train to get the best and include: S1-2-1: Establish the loss function value according to the following formula: , Where, is the cumulative value of all negative emotional information before time t, ; S1-2-2: Update K according to the following formula; , Where, is the adjustment coefficient, is the gradient of K; S1-2-3: Update r according to the following formula: , Where, is the adjustment coefficient, is the gradient of r; S1-2-4: Determine whether the loss function value L is the minimum. If so, output the optimal K and r respectively. 、 , and then execute step S1-2-5; if the loss function value, L does not reach the minimum, so that the changed K, r are replaced by the values ​​before the change and then repeat steps S1-2-1 to S1-2-4; S1-2-5: Conduct negative emotion guidance at time t0, and establish a negative emotion dynamic evolution model for the period after time t0 to predict the trend of negative emotions for a period after time t0: .

[0016] The dynamic evolution model of negative emotions before time t0 is: .

[0017] The method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention also includes: S3: Classify events that cause negative emotions according to the following rules: like and , then the events that cause negative emotions are general events; like and , then the event that causes negative sentiment is a major event; If and , then the event that causes negative sentiment is an extremely major event. In the formula, , , , are thresholds. For example, , , , .

[0018] In the present invention, obtaining the critical time nodes of the calculation evolution stage by using the negative sentiment dynamic evolution model includes: S2-1: Performing a third-order derivative on the negative sentiment dynamic evolution model established after time t0, and letting to obtain the optimal guidance time windows [0, t1], [t1, t2], [t2, , , , where the time window [0, t1] is the incubation period, [t1, t2] is the outbreak period, and [t2, is the stable period. It should be noted that although the present invention introduces negative sentiment guidance at time t0, when calculating the optimal guidance time window, the model is modified to , and then a third-order derivative is performed to obtain the optimal time window.

[0019] The public opinion intelligent deduction and guidance method based on evolutionary modeling provided by the present invention further includes: S4: Adopting different strategies according to different scenarios: For general events, adopt the "priority-degree-then-time" strategy, that is, first optimize the guidance coefficient α, and then optimize the negative sentiment guidance time t0; for extremely major events, adopt the "priority-time-then-degree" strategy: force t0 < t1, and then dynamically adjust the guidance coefficient α; for major events, the "priority-degree-then-time" strategy or the "priority-time-then-degree" strategy can be adopted.

[0020] By establishing a negative sentiment dynamic evolution model, the present invention transforms the guidance timing of subjective experience judgment into quantitative calculation of time nodes based on a mathematical model, accurately divides the public opinion evolution stages: the incubation period, the outbreak period, and the stable period, and provides a quantitative basis for the selection of the guidance timing. The calculation of the critical time nodes t1 and t2 directly supports the optimization of the timeliness of the strategy. The classification of the scenario strategies avoids secondary risks caused by a unified strategy (such as the public opinion rebound caused by the post-guidance of extremely major events), improves the safety of the strategy, and solves the problem of scenario adaptation.

[0021] The method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention further includes: S5: Conducting temporal-effectiveness collaborative simulation deduction, specifically including: S5-1: Determining the combined scenarios of the starting time t0 of the negative emotion dynamic evolution model and the guidance coefficient α. In the present invention, t0 < t1 is pre-guidance, and t0 > t1 is post-guidance; S5-2: According to the negative emotion dynamic evolution model, pre演 the evolution trajectory of the negative emotion information volume under different strategies, and output key indicators: peak suppression rate (for example, when α = 2, the peak decreases by 66.67%), and the acceleration duration of the stable period. The acceleration duration of the stable period of pre-guidance is shortened by 50% compared with that of post-guidance; S5-3: Designing a strategy priority rule: In major and extremely important events, if t0 > t1, automatically trigger a secondary public opinion warning.

[0022] Quantitatively predict the strategy effect through temporal-effectiveness collaborative simulation deduction, avoid the traditional trial-and-error cost, and solve the problem of unpredictable effects. Reverse-optimize the strategy parameters according to the simulation results to form a "modeling-deduction-feedback" closed loop.

[0023] The method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention further includes: S6: Visualizing the intelligent guidance effect of public opinion, specifically including designing a multi-dimensional visualization panel to dynamically display the real-time public opinion situation, such as the information volume curve and the emotional polarity distribution; comparing the simulation deduction results, that is, the effect differences of different combinations of α and t0; and conducting a risk warning for strategy execution.

[0024] Provide intuitive strategy optimization basis for decision-makers by visualizing the intelligent guidance effect of public opinion and reduce the technical threshold.

[0025] Before step S1, the method for intelligent deduction and guidance of public opinion based on evolutionary modeling provided by the present invention further includes: Using a clustering algorithm to cluster network public opinion texts and separating negative emotion texts, specifically including: S3-1: Establish a generative artificial intelligence model based on network public opinion texts, and generate K words according to the generative artificial intelligence model; S3-2: Cluster the K words and separate the words representing negative emotions.

[0026] In the present invention, the construction process of the generative artificial intelligence model includes: S3-1-1: Generating a topic distribution for the d-th document of the network public opinion text: , where is the hyperparameter of the d-th topic distribution, d = 1,..., D, and D is a positive integer greater than or equal to 1; S3-1-2: Generating a word distribution for the k-th topic of the network public opinion text : , Where, is the hyperparameter of the kth topic distribution, k=1,…,K, where K is a positive integer greater than or equal to 1; represents Dirichlet distribution; S3-1-3: Topic distribution generated from the dth document Sampling a topic : , Where, Represents a multinomial distribution, e=1,…,N d ; S3-1-4: From the subject Corresponding word distribution Sample a word : ; S3-1-5: Generate joint probability distribution according to the following formula : ; S3-1-6: Use the following generative AI model to infer the topic distribution θ of documents, the word distribution ϕ of topics, and the topic distribution z of words: , where , Where, is the posterior distribution, For a simple distribution, are the two parameters of a simple distribution; represents the antiphase divergence; represents the expected posterior distribution of word w for , represents the prior about word w; S3-1-7: Generate K words based on the topic distribution θ, the word distribution ϕ of the topic, and the topic distribution z of the word.

[0027] The present invention also provides a system comprising a storage medium and one or more processors, wherein the storage medium is used to store code for compiling part or all of the steps in any of the above-mentioned methods for intelligent deduction and guidance of public opinion based on evolutionary modeling into a computer program using a computer language, and the computer program can be called and executed by one or more processors.

[0028] A new technical solution formed by a combination of one or more of the above steps disclosed in the present invention also falls within the scope of the present invention.

[0029] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent deduction and guidance of public opinion based on evolutionary modeling, characterized by: The steps include: S1: Establish a negative sentiment dynamic evolution model based on negative sentiment texts; S2: Use the negative emotion dynamic evolution model to obtain the critical time nodes of the evolution stage.

2. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 1 is characterized in that: The negative sentiment dynamic evolution model established based on negative sentiment texts includes: S1-1: Building a regression model for: ， Where, is the growth rate, K is the information limit; It is the amount of negative emotional information at the initial moment of negative emotion; S1-2: Regression Model Train to get the best and .

3. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 2 is characterized in that: For regression models Train to get the best and include: S1-2-1: Establish the loss function value according to the following formula: ， Where, is the cumulative value of all negative emotional information before the moment, and the time series of the cumulative value of information is obtained. ; S1-2-2: Update K according to the following formula; ， Where, is the adjustment coefficient, is the gradient of K; S1-2-3: Update r according to the following formula: ， Where, is the adjustment coefficient, is the gradient of r; S1-2-4: Determine whether the loss function value L is the minimum. If so, output the optimal K and r respectively. 、 , and then execute step S1-2-5; if the loss function value, L does not reach the minimum, so that the changed K, r are replaced by the values ​​before the change and then repeat steps S1-2-1 to S1-2-4; S1-2-5: Conduct negative emotion guidance at time t0, and establish a negative emotion dynamic evolution model for the period after t0 to predict the trend of negative emotions for a period of time after t0: 。 4. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 3 is characterized in that: Also includes: S3: Classify events that cause negative emotions according to the following rules: like and , then the events that cause negative emotions are general events; like and , then the event that caused negative emotions was a major event; like and , then the event that caused negative emotions is a particularly serious event, Where, 、 、 、 is the threshold.

5. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 4 is characterized in that: The critical time nodes of the evolution stage calculated by using the negative emotion dynamic evolution model include: S2-1: Take the third-order derivative of the negative emotion dynamic evolution model established after t0, and let Get the best guidance time window [0,t1], [t1,t2], [t2, ], , .

6. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 1 is characterized in that: Before step S1, the method further includes: clustering the online public opinion texts using a clustering algorithm to separate the negative sentiment texts.

7. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 6 is characterized in that: Clustering algorithms are used to cluster online public opinion texts, separating out negative sentiment texts including: S3-1: Establish a generative artificial intelligence model based on online public opinion text, and generate K words based on the generative artificial intelligence model; S3-2: Cluster the K words and separate the words that express negative emotions.

8. The method for intelligent deduction and guidance of public opinion based on evolutionary modeling according to claim 7 is characterized in that: The process of building a generative AI model includes: S3-1-1: Generate topic distribution for the dth document of online public opinion text: ， Where, is the hyperparameter of the d-th topic distribution, d=1,…,D, where D is a positive integer greater than or equal to 1; S3-1-2: Generate word distribution for the kth topic of online public opinion text : ， Where, is the hyperparameter of the kth topic distribution, k=1,…,K, where K is a positive integer greater than or equal to 1; represents Dirichlet distribution; S3-1-3: Topic distribution generated from the dth document Sampling a topic : ， Where, Represents a multinomial distribution, e=1,…,N d ; S3-1-4: From the subject Corresponding word distribution Sample a word : ; S3-1-5: Generate joint probability distribution according to the following formula : ； S3-1-6: Use the following generative AI model to infer the topic distribution θ of documents, the word distribution ϕ of topics, and the topic distribution z of words: ， Where, , Where, is the posterior distribution, For a simple distribution, are the two parameters of a simple distribution; represents the antiphase divergence; represents the expected posterior distribution of word w for , represents the prior about word w; S3-1-7: Generate K words based on the topic distribution θ, the word distribution ϕ of the topic, and the topic distribution z of the word.

9. A system comprising a storage medium and one or more processors, wherein the storage medium is used to store code that uses a computer language to compile some or all of the steps in the method for intelligent deduction and guidance of public opinion based on evolutionary modeling as described in any one of claims 1-8 into a computer program, and the computer program can be called and executed by one or more processors.

Images