Social network user emotion propagation model based on game theory

Abstract

The invention discloses a social network user emotion propagation model based on a game theory, and aims to overcome the technical defect that an emotion propagation model cannot accurately describe a real emotion propagation process in the existing emotion analysis and propagation simulation work in the prior art. According to the technical scheme provided by the invention, a model construction method of the social network user emotion propagation model based on the game theory comprises the following steps: performing emotion analysis on a target user node according to a preset emotion analysis method, and extracting an emotion label of the node; constructing an emotional game income matrix according to the emotional label of each node; adjusting the basic propagation probability in the SIR infectious disease model according to the emotional game income matrix; according to the social enhancement effect, the basic immune probability in the SIR infectious disease model is adjusted; and adjusting the structure of the SIR infectious disease model according to a forgetting mechanism to obtain a social network user emotion propagation model. The method is suitable for sentiment analysis and propagation simulation work.

Description

Technical Field

[0001] The invention relates to the technical field of social network analysis, and in particular to a social network user emotion propagation model based on game theory. Background Art

[0002] In the field of social network analysis, the study of sentiment communication has received increasing attention in recent years. With the rapid development of the Internet and social media, various online platforms have gradually replaced traditional media and become the main channels for people to express their emotions and share information. Through these platforms, users can communicate and express their emotions and opinions in real time, thus forming a complex sentiment communication dynamic on social networks. Existing research mainly focuses on the following aspects:

[0003] First, text-based sentiment analysis methods have been widely used. Researchers use natural language processing technology, dictionaries and machine learning models to extract emotional tendencies in user-posted content. For example, the sentiment dictionary method is widely used. By constructing a dictionary including positive and negative sentiment words, the sentiment of text in social networks is classified and scored to analyze the dynamic characteristics of sentiment propagation. In addition, machine learning models, especially deep learning technology, are also used to train sentiment classifiers to achieve higher sentiment recognition accuracy.

[0004] Secondly, the construction of emotion propagation models is also an important direction of current research. Commonly used models include the SIR (susceptible-infectious-recovered) model and its variants. These models were originally used to describe the spread of infectious diseases, but have also been introduced into emotion propagation research to simulate the spread of emotions in social networks. However, the traditional SIR model mainly focuses on the calculation of propagation probability and does not fully consider the complexity of emotional tendencies and their propagation decisions between individuals, which leads to a certain deviation between the simulation results of emotion propagation and the actual situation.

[0005] Although existing research has made a lot of efforts in sentiment analysis and communication simulation, it often ignores the decision-making factors of individuals in the process of sentiment communication. For example, when individuals decide whether to spread a piece of information, they are not only affected by the emotional tendency of the information itself, but also consider the emotional similarity with other users and the potential benefits or costs of communication. This decision-making process has the nature of a game, and traditional models lack the description of this game behavior, which limits the applicability of the model in complex social environments.

[0006] Therefore, the problem existing in the prior art is that there is a lack of consideration of the emotional game between users, which results in the inability of the emotional communication model to accurately portray the real emotional communication process, especially when individual communication decisions depend on interest games. Summary of the invention

[0007] In order to solve the technical defect that the existing emotion analysis and communication simulation work lacks consideration of the emotional game between users, resulting in the emotion communication model being unable to accurately depict the real emotion communication process, the technical solution provided by the present invention is:

[0008] A social network user sentiment propagation model based on game theory, the model construction method includes:

[0009] The step of performing sentiment analysis on the target user node according to a preset sentiment analysis method and extracting the sentiment label of the node;

[0010] The step of constructing an emotional game payoff matrix according to the emotional labels of each node;

[0011] The step of adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix;

[0012] A step of adjusting the basic immunity probability in the SIR epidemic model according to the social reinforcement effect;

[0013] The structure of the SIR epidemic model is adjusted according to the forgetting mechanism to obtain the step of the social network user emotion propagation model.

[0014] Furthermore, a preferred implementation is provided to construct an emotional game benefit matrix based on the emotional labels of each node and the relationship between the emotional tendencies of the information communicator and the emotional tendencies of the information receiver.

[0015] Further, a preferred implementation is provided, wherein the preset sentiment analysis method is specifically:

[0016] Establish a sentiment dictionary and assign different sentiment scores or labels with different sentiment tendencies to the sentiment words in the dictionary.

[0017] Furthermore, a preferred embodiment is provided, wherein the adjustment of the immunity probability in the social reinforcement effect is based on the proportion of the immunity status of the surrounding nodes, and the adjustment amplitude is adjusted by a control parameter.

[0018] Furthermore, a preferred implementation is provided, in which the propagation probability adjustment in the SIR model is based on the calculation result of the game equilibrium point in the game profit matrix between the propagation node and the receiving node.

[0019] Furthermore, a preferred implementation is provided, wherein the calculation of the game equilibrium point is based on a Bayesian Nash equilibrium condition.

[0020] Based on the same inventive concept, the present invention also provides a device for constructing a social network user emotion propagation model based on game theory, comprising:

[0021] A module that performs sentiment analysis on target user nodes according to a preset sentiment analysis method and extracts the sentiment labels of the nodes;

[0022] A module for constructing an emotional game payoff matrix according to the emotional labels of each node;

[0023] A module for adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix;

[0024] A module for adjusting the basic immunity probability in the SIR infectious disease model according to the social reinforcement effect;

[0025] The structure of the SIR infectious disease model is adjusted according to the forgetting mechanism to obtain a module of the social network user emotion propagation model.

[0026] Based on the same inventive concept, the present invention also provides a computer storage medium for storing a computer program, and when the computer reads the computer program, the computer executes the described method.

[0027] Based on the same inventive concept, the present invention also provides a computer, including a processor and a storage medium. When the processor reads a computer program stored in the storage medium, the computer executes the method described.

[0028] Based on the same inventive concept, the present invention also provides a computer program product, which is a computer program. When the computer program is executed, the method described above is implemented.

[0029] Compared with the prior art, the technical solution provided by the present invention is beneficial in that:

[0030] The introduction of the emotional game payoff matrix allows the model to more accurately characterize the trade-off between the benefits and costs of information dissemination during the dissemination process. This payoff matrix takes into account the emotional consistency and inconsistency between the dissemination node and the receiving node, and quantifies the dissemination benefits in different situations. Compared with traditional models, it can more realistically simulate the dissemination decisions of users in complex social networks.

[0031] The introduction of social reinforcement effect improves the model's simulation effect of propagation in complex networks. By dynamically adjusting the number of immune nodes around a node, the model can better describe the mutual influence between individuals during the process of emotional propagation. This mechanism makes the adjustment of propagation probability more consistent with the law of emotional propagation in reality, especially when social connections between individuals have an important impact on emotional propagation.

[0032] The application of the forgetting mechanism effectively improves the accuracy and stability of the model for sentiment propagation. In the traditional SIR model, nodes in the infected state will remain active all the time. However, after the introduction of the forgetting mechanism, the nodes will forget the received information with a certain probability and turn to the immune state. This mechanism makes the model closer to the changing process of user emotions in reality and can better describe the attenuation effect of sentiment propagation, especially in the process of negative emotions gradually fading.

[0033] Compared with existing models that only rely on transmission probability and infection state transition, this scheme significantly improves the accuracy and applicability of the model in terms of emotion transmission by introducing emotional game, social reinforcement effect and forgetting mechanism. These mechanisms work together to enable the model to not only more accurately predict the transmission path of emotions in the network, but also better reflect the emotional transmission behavior of individuals in social networks.

[0034] Suitable for use in sentiment analysis and communication simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flowchart of a method for constructing a social network user sentiment propagation model based on game theory;

[0036] Figure 2 Schematic diagram of the adjustment of the SIR infectious disease model structure. DETAILED DESCRIPTION

[0037] In order to make the advantages and benefits of the technical solution provided by the present invention more clearly reflected, the technical solution provided by the present invention is now further described in detail in conjunction with the accompanying drawings, specifically:

[0038] Implementation method 1: This implementation method provides a social network user emotion propagation model based on game theory, and the model construction method includes:

[0039] The step of performing sentiment analysis on the target user node according to a preset sentiment analysis method and extracting the sentiment label of the node;

[0040] The step of constructing an emotional game payoff matrix according to the emotional labels of each node;

[0041] The step of adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix;

[0042] A step of adjusting the basic immunity probability in the SIR epidemic model according to the social reinforcement effect;

[0043] The structure of the SIR epidemic model is adjusted according to the forgetting mechanism to obtain the step of the social network user emotion propagation model.

[0044] Specifically, they include:

[0045] Step 1: Sentiment analysis and label extraction

[0046] First, the improved dictionary-based sentiment analysis method is used to perform sentiment analysis on user nodes in social networks. Specifically, a sentiment dictionary is constructed, which includes commonly used word segmentation dictionaries, stop word dictionaries, and sentiment word dictionaries. In order to improve the accuracy of sentiment analysis, 7 degree word dictionaries, 1 negation word dictionary, and 2 sentiment word dictionaries (positive and negative sentiment word dictionaries, respectively) are also introduced. During the analysis process, the text content of the user node is segmented, and after removing stop words, the sentiment words in the text are found. The sentiment words are matched based on the sentiment dictionary and the sentiment score is calculated. The sentiment tendency of the node is determined based on the label with the highest score or the largest number, and the output is the sentiment label of the node.

[0047] In this process, each text is analyzed according to different sentiment words in the dictionary, and different degree words weight the sentiment intensity of the sentiment words. For example, "very like" has a greater sentiment intensity than "like", so it is given a higher weight in the analysis process. Negative words also modify the tendency of sentiment words. For example, "dislike" will reverse the sentiment of "like" to negative sentiment. Through these steps, the results of sentiment analysis can more accurately reflect the user's true emotional state.

[0048] Step 2: Construct the emotional game benefit matrix

[0049] Based on the user node emotional labels extracted in step 1, an emotional game benefit matrix is ​​constructed. The benefit matrix takes into account the different situations of emotional consistency and inconsistency between the propagation node and the receiving node, and quantifies the propagation benefits in each situation. For example, when the emotional tendencies of the propagation node and the receiving node are consistent and both choose to propagate, both parties can obtain higher benefits, while when the emotional tendencies are inconsistent, the propagation node may face penalties. The output of this step is the emotional game benefit matrix, which is used to model the propagation process in the subsequent steps.

[0050] When constructing the game payoff matrix, it is necessary to consider node selection in various scenarios. For example, nodes can choose to spread or not spread information, and receiving nodes can choose to accept or not receive information. When the emotions match, both spreading and receiving bring positive benefits, while when the emotions do not match, spreading may bring penalties or negative benefits. This payoff setting can simulate the decision-making behavior of individuals in social networks during the process of emotional communication, thereby more accurately predicting the communication results.

[0051] Step 3: Integrate the emotional game payoff matrix and the SIR model

[0052] The emotional game payoff matrix constructed in step 2 is integrated with the SIR infectious disease model. The basic propagation probability in the SIR model is dynamically adjusted through the emotional game payoff matrix, so that the propagation probability is not only affected by the node state, but also depends on the emotional game relationship between the propagation node and the receiving node. Specifically, the propagation probability is optimized and adjusted using the Bayesian Nash equilibrium condition to better simulate the propagation process of emotions in the network.

[0053] In this process, the emotional state of the transmission node and the receiving node and the benefit matrix determine the probability of information transmission. If the benefits brought by the transmission are high, the transmission probability will increase accordingly, and vice versa. By introducing game theory, the model can reflect the rational choices of individuals in the transmission process and simulate the transmission dynamics that are closer to the real situation.

[0054] Step 4: Introduce social reinforcement effect to adjust the probability of immunity

[0055] Based on the SIR model, the social reinforcement effect is introduced to adjust the basic immunity probability. By considering the number of immune nodes around the node, the immunity probability of the node is dynamically adjusted. Specifically, as the number of immune nodes around increases, the immunity probability of the node will also increase accordingly. When there are no immune nodes around, the immunity probability is equal to the basic immunity probability. This dynamic adjustment method makes the model closer to the immune behavior of individuals in social networks in reality.

[0056] The social reinforcement effect can describe the herd behavior of individuals in social networks. When there are a large number of immune nodes around a node, the probability that the node will choose not to spread or be immune will also increase. This mechanism can better describe the phenomenon of emotional transmission inhibition caused by social pressure or herd mentality in reality, thereby improving the authenticity and accuracy of the model.

[0057] Step 5: Introduce the forgetting mechanism to adjust the SIR model

[0058] Introducing the forgetting mechanism into the traditional SIR model makes the nodes in the infected state forget the received information and change to the immune state with a certain probability. As the forgetting probability increases, the number of infected nodes that become immune nodes will increase, which effectively describes the attenuation effect of emotion propagation. The introduction of the forgetting mechanism enables the model to better simulate the emotional decay process in reality after the emotional propagation ends, which is especially suitable for the gradual decay of negative emotions.

[0059] The introduction of the forgetting mechanism enables the model to dynamically simulate the end process of emotional communication. In real social networks, negative emotions tend to fade away over time, and this phenomenon is reflected by the forgetting mechanism. As the probability of forgetting changes, the model can describe the persistence and fading speed of emotions in different situations, thus providing a theoretical basis for managing and intervening in the spread of negative emotions.

[0060] Implementation method 2: This implementation method further limits the social network user emotion propagation model based on game theory provided in implementation method 1. An emotion game benefit matrix is ​​constructed according to the emotion label of each node and the relationship between the emotion tendency of the information communicator and the emotion tendency of the information receiver.

[0061] Implementation method 3: This implementation method further limits the social network user emotion propagation model based on game theory provided in implementation method 1. The preset emotion analysis method is specifically:

[0062] Establish a sentiment dictionary and assign different sentiment scores or labels with different sentiment tendencies to the sentiment words in the dictionary.

[0063] Implementation method 4: This implementation method is a further limitation of the social network user emotion propagation model based on game theory provided in implementation method 1. The adjustment of the immunity probability in the social reinforcement effect is based on the proportion of the immune status of the surrounding nodes, and the adjustment range is adjusted by controlling parameters.

[0064] Implementation method five: This implementation method is a further limitation of a social network user emotion propagation model based on game theory provided in implementation method one. The propagation probability adjustment in the SIR model is based on the calculation result of the game equilibrium point in the game profit matrix between the propagation node and the receiving node.

[0065] Implementation method 6: This implementation method further limits the social network user emotion propagation model based on game theory provided in implementation method 1, and the calculation of the game equilibrium point is based on the Bayesian Nash equilibrium condition.

[0066] Embodiment 7: This embodiment provides a device for constructing a social network user emotion propagation model based on game theory, including:

[0067] A module that performs sentiment analysis on target user nodes according to a preset sentiment analysis method and extracts the sentiment labels of the nodes;

[0068] A module for constructing an emotional game payoff matrix according to the emotional labels of each node;

[0069] A module for adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix;

[0070] A module for adjusting the basic immunity probability in the SIR infectious disease model according to the social reinforcement effect;

[0071] The structure of the SIR infectious disease model is adjusted according to the forgetting mechanism to obtain a module of the social network user emotion propagation model.

[0072] Embodiment 8: This embodiment provides a computer storage medium for storing a computer program. When the computer reads the computer program, the computer executes the method provided in embodiment 1.

[0073] Embodiment 9: This embodiment provides a computer, including a processor and a storage medium. When the processor reads a computer program stored in the storage medium, the computer executes the method provided in embodiment 1.

[0074] Embodiment 10: This embodiment provides a computer program product, which is a computer program. When the computer program is executed, the method provided in embodiment 1 is implemented.

[0075] Implementation Method XI: Combination Figure 1 and 2 This embodiment is described in detail by using specific examples to further describe the technical solution provided above. Specifically:

[0076] Reference Figure 1 As shown, this embodiment proposes an embodiment of a social network user emotion propagation model based on game theory, and the construction method includes the following steps:

[0077] Step 1: Use the improved dictionary-based sentiment analysis method to perform sentiment analysis on user nodes and extract the sentiment labels of nodes;

[0078] Step 2: Based on the emotional labels of each node and the relationship between the emotional tendencies of the information communicator and the information receiver, the emotional game payoff matrix is ​​constructed;

[0079] Step 3: Integrate the emotional benefit game matrix with the SIR infectious disease model, and adjust the basic transmission probability in the infectious disease model through the emotional game benefit matrix;

[0080] Step 4: Use the social reinforcement effect to adjust the basic immunity probability in the transmission model;

[0081] Step 5: Use the forgetting mechanism to adjust the SIR infectious disease model structure and complete the construction of the social network user emotion propagation model based on game theory.

[0082] In step one, the improved dictionary-based sentiment analysis method is used to perform sentiment analysis on the user node. Specifically, the dictionary-based sentiment analysis method first establishes a sentiment dictionary. When constructing the sentiment dictionary, in addition to the commonly used word segmentation dictionary, stop word dictionary and sentiment word dictionary, 7 degree word dictionaries, 1 negation word dictionary and 2 sentiment word dictionaries (one positive sentiment word dictionary and one negative sentiment word dictionary respectively) are added to increase the accuracy of sentiment analysis. Different sentiment scores are assigned to the sentiment words in the dictionary or they are labeled with different sentiment tendencies. Then, in the process of sentiment analysis, the text of the user node is segmented, and the stop words in the text are removed according to the word segmentation dictionary and the stop word dictionary to find the sentiment words in the text. The sentiment words in the text are directly matched with the sentiment words in the sentiment dictionary. According to the sentiment scores or sentiment tendency labels of all words in the text, the label with the highest score or the largest number is selected as the final sentiment tendency of the text.

[0083] Specifically, when performing sentiment analysis on the sentence "I particularly like the city of Wuhan! Because Wuhan has many beautiful attractions. However, I don't like the weather in Wuhan, because the weather in Wuhan is a bit bad, and it makes people feel uncomfortable when it is hot." To analyze the sentiment tendency of a sentence, we must first find the sentiment words, degree adverbs, and negative words in the sentence. The positive sentiment words in the sentence are: "good-looking" and "like", and the negative sentiment words are: "bad" and "unpleasant". Each time a positive word appears, the score is +1, and each time a negative word appears, the score is -1. The sentiment words "like", "good-looking" and "bad" are preceded by a degree adverb. "Excellent" is more emotional than "relatively good" and "good", and "too bad" is also more emotional than "a bit bad". Therefore, it is necessary to look for degree adverbs after finding the sentiment words (different degree adverbs have different weights here). We divide the degree words into 6 types, namely: most, very, more, ish, insufficient and over. Through test calculation, the scores of the above 6 emotional words are 4, 3, 2, 0.5, -0.3 and -0.5 respectively. "I particularly like the city of Wuhan" is followed by an exclamation mark, which means strong emotions. Therefore, it is found that the exclamation mark can be +2 for the emotional score. And when finding emotional words, you need to look forward for negative words. For example, the words "no" and "can't". In addition, the number of occurrences of these negative words must be counted. If it is a singular number, the emotional score is multiplied by -1, but if it is an even number, the emotion is not reversed. In this sentence, it can be seen that there is only one "no" in front of "like", so the emotional value of "like" should be reversed and multiplied by -1. In the method adopted in this embodiment, positive emotions and negative emotions are calculated separately, because in a text, there are praises and criticisms, and one score cannot be used to represent its emotional tendency. By scoring each sentence and finally weighted summing, a final positive score and negative score are obtained. Whichever score is larger, the emotion of the text tends to that side. In this example, the final scores of the five sentences are [positive score, negative score] = [8, 1.5], 8>1.5, so the sentiment of the whole sentence is positive. Therefore, sentiment analysis can be performed on the text of the node to extract the sentiment label.

[0084] In step 2, the emotional game benefit matrix is ​​constructed. The specific construction method is:

[0085] Game payoff matrix

[0086]

[0087] Among them, the cost of a propagation node to propagate to a node is c; the benefits of the two nodes when a piece of information that matches the sentiment is successfully propagated is x; the penalty c is incurred when the propagation node successfully propagates a piece of information that does not match the sentiment; the benefits of the two nodes when they prevent a piece of information that does not match the sentiment from propagating are d. When the sentiment tendencies between the propagation node and the receiving node match, if the propagation node chooses to propagate and the receiving node chooses to receive, the benefit of the receiving node is x, and the benefit of the propagation node is xc; when the sentiment tendencies between the propagation node and the receiving node match, if the propagation node chooses to propagate and the receiving node chooses not to receive, the benefit of the receiving node is 0, and the benefit of the propagation node is -c; when the sentiment tendencies between the propagation node and the receiving node match, if the propagation node chooses not to propagate, the benefits of both the receiving node and the propagation node are 0; when the sentiment tendencies between the propagation node and the receiving node do not match, if the propagation node chooses to propagate and the receiving node chooses to receive , the benefit of the receiving node is -c, and the benefit of the propagation node is -2c; when the sentiment tendency between the propagation node and the receiving node is inconsistent, if the propagation node chooses to propagate and the receiving node chooses not to receive, the benefit of the receiving node is d, and the benefit of the propagation node is -2c; when the sentiment tendency between the propagation node and the receiving node is inconsistent, if the propagation node chooses not to propagate and the receiving node chooses to receive, the benefit of the receiving node is 0, and the benefit of the propagation node is dc; when the sentiment tendency between the propagation node and the receiving node is inconsistent, if the propagation node chooses not to propagate and the receiving node chooses not to receive, the benefit of the receiving node is d, and the benefit of the propagation node is dc. In step three, the basic propagation probability in the infectious disease model is adjusted using the emotional game benefit matrix. The specific adjustment method is:

[0088] p·x=(1-p)·d

[0089] The conditions for the existence of the Bayesian Nash equilibrium of the game payoff matrix are x>c>0 and dc>0. Its Bayesian Nash equilibrium is shown in the above formula, which can be converted into the following formula:

[0090]

[0091] Where p is the adjusted propagation probability, x is the benefit of the two nodes when a piece of information that matches the sentiment is successfully propagated, and d is the benefit of preventing the propagation of a piece of information that does not match the sentiment.

[0092] In step 4, the social reinforcement effect is used to adjust the basic immunity probability in the transmission model. The specific adjustment method is:

[0093] γ=γ base ·(log a( b·n)+1)

[0094] Among them, γ is the adjusted immunity probability, γbase is the basic immunity probability of the propagation model, n is the number of immunity nodes around the node, a and b are control parameters. By controlling the values ​​of a and b parameters, the degree of adjustment of the immunity probability can be controlled; as the number of surrounding immunity nodes n increases, the immunity probability γ will increase on the basis of the basic immunity probability γbase. When n=0, the immunity probability is equal to the basic immunity probability.

[0095] In step 5, the forgetting mechanism is used to adjust the structure of the SIR epidemic model, as follows: Figure 2 shown.

[0096] A forgetting mechanism is introduced on the basis of the traditional SIR model. The nodes in the infected state will forget the information they have received due to the forgetting mechanism and thus change to the immune state. As the forgetting probability increases, the number of infected nodes that forget to become immune nodes will increase. Correspondingly, the number of immune nodes after the propagation is completed will be more than the number of immune nodes in the model without the introduction of the forgetting mechanism; as the forgetting probability decreases, the number of infected nodes that forget to become immune nodes will increase. Correspondingly, the number of immune nodes after the propagation is completed will be more than the number of immune nodes in the model without the introduction of the forgetting mechanism, but the number of immune nodes added is reduced compared to when the forgetting probability is greater.

[0097]

[0098] The above formula is the propagation dynamics formula of the social network user emotion propagation model based on game theory, where β is the propagation probability, γ is the immunity probability, and η is the forgetting probability. Therefore, when β is larger, the probability of infection from S state to I state is greater, and the more S nodes become I state. Therefore, when γ is larger, the probability of immunity from I state to R state is greater, and the more I nodes become R state.

[0099] The technical solution provided by the present invention is further described in detail above through several specific implementation modes in order to highlight the advantages and benefits of the technical solution provided by the present invention. However, the several specific implementation modes described above are not intended to be used as limitations on the present invention. Any reasonable modification and improvement of the present invention, combination of implementation modes and equivalent substitution within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A social network user sentiment propagation model based on game theory, characterized in that: Model building methods include: The step of performing sentiment analysis on the target user node according to a preset sentiment analysis method and extracting the sentiment label of the node; The step of constructing an emotional game payoff matrix according to the emotional labels of each node; The step of adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix; A step of adjusting the basic immunity probability in the SIR epidemic model according to the social reinforcement effect; The structure of the SIR epidemic model is adjusted according to the forgetting mechanism to obtain the step of the social network user emotion propagation model.

2. According to the game theory-based social network user emotion propagation model according to claim 1, it is characterized in that: According to the emotional labels of each node and the relationship between the emotional tendency of the information communicator and the emotional tendency of the information receiver, an emotional game benefit matrix is ​​constructed.

3. The game theory-based social network user emotion propagation model according to claim 1, characterized in that: The preset sentiment analysis method is specifically: Establish a sentiment dictionary and assign different sentiment scores or labels with different sentiment tendencies to the sentiment words in the dictionary.

4. The game theory-based social network user sentiment propagation model according to claim 1, characterized in that: The adjustment of the immunity probability in the social reinforcement effect is based on the proportion of the immunity status of the surrounding nodes, and the adjustment amplitude is adjusted by the control parameter.

5. The game theory-based social network user sentiment propagation model according to claim 1, characterized in that: The propagation probability adjustment in the SIR model is based on the calculation result of the game equilibrium point in the game profit matrix between the propagation node and the receiving node.

6. The game theory-based social network user sentiment propagation model according to claim 1, characterized in that: The calculation of the game equilibrium point is based on the Bayesian Nash equilibrium condition.

7. A device for constructing a social network user emotion propagation model based on game theory, characterized in that: include: A module that performs sentiment analysis on target user nodes according to a preset sentiment analysis method and extracts the sentiment labels of the nodes; A module for constructing an emotional game payoff matrix according to the emotional labels of each node; A module for adjusting the basic transmission probability in the SIR infectious disease model according to the emotional game payoff matrix; A module for adjusting the basic immunity probability in the SIR infectious disease model according to the social reinforcement effect; The structure of the SIR infectious disease model is adjusted according to the forgetting mechanism to obtain a module of the social network user emotion propagation model.

8. A computer storage medium for storing a computing program, characterized in that: When the computer reads the computer program, the computer executes the method of claim 1.

9. A computer, comprising a processor and a storage medium, characterized in that: When the processor reads the computer program stored in the storage medium, the computer executes the method of claim 1 .

10. A computer program product, being a computer program, characterized in that When the computer program is executed, the method of claim 1 is implemented.

Images