A method for detecting the spread of public opinion

Abstract

The present invention belongs to the field of information dissemination and control technology, and specifically relates to a method for detecting the dissemination of public opinion. The present invention mainly defines attributes of users in a social network, and respectively assigns attributes of influence I and susceptibility S, where influence refers to the ability of a user to influence other users, and susceptibility refers to the degree of influence by other users, and then obtains the influence and susceptibility of all users in the social network through iteration, thereby monitoring high-ranking users instead of monitoring all users, thereby improving efficiency and saving resources.

Description

Technical Field

[0001] The present invention belongs to the technical field of information dissemination and control, and specifically relates to a method for detecting the dissemination of public opinion. Background Art

[0002] Public opinion control in social networking platforms involves social stability, public interest protection, personal privacy, and brand reputation. The detection of public opinion helps prevent the spread of false information, rumors, and hate speech, maintain social order, and reduce conflicts. At the same time, the detection of public opinion can also protect the public interest and prevent harmful information from causing harm to public health and national security. In addition, online public opinion control also plays a role in protecting personal privacy and corporate brand reputation. By detecting the spread of untrue or negative information, it ensures that personal privacy information is not abused and the reputation of the company is not damaged. The main methods of controlling public opinion on social platforms include batch blocking and deleting posts of accounts that participate in forwarding public opinion information. This type of method has the problems of being unable to grasp the blocking strength, excessively interfering with users' privacy information, untimely public opinion control, and affecting the normal social experience of ordinary users. Summary of the invention

[0003] In view of the above problems, the present invention proposes a method for detecting the spread of public opinion.

[0004] The technical solution of the present invention is:

[0005] A method for predicting the spread of public opinion includes the following steps:

[0006] The users in the social network are defined by attributes, and the attributes influence I and susceptibility S are given respectively, where influence refers to the ability of a user to influence other users, and susceptibility refers to the degree of influence by other users; thus, the probability p_ij that the behavior of user i affects user j is associated with the influence of user i and the susceptibility of user j, expressed as p ij =I i S j ;

[0007] Collect historical interaction data of users within a set time period through social platforms and estimate the interaction probability between users ω ij ;

[0008] Iterative calculation of user influence and susceptibility:

[0009]

[0010]

[0011] Among them, f i =∑ j A ij ωij ,g j =∑ i A ij ω ij ,A ij is an element in the adjacency matrix A. When A ij =1, indicating that user j has interacted with the information posted by user i at least once, otherwise user j has not interacted with the information posted by user i; when iterates to When the value of is less than the set threshold, the iteration stops, where N is the number of users in the network and t is the number of algorithm iterations;

[0012] Through iteration, the influence and susceptibility of all users in the social network are obtained, and the users ranked in the top 1% in susceptibility are marked. When online public opinion spreads, the marked users are monitored and the information they post on the social network is obtained to predict and monitor the spread of public opinion.

[0013] The beneficial effects of the present invention are as follows: highly susceptible users are calculated through historical interaction data, and highly susceptible users who are highly correlated with public opinion are detected, thereby avoiding monitoring of all users in the network, improving detection efficiency, and saving computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0015] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0016] like Figure 1 As shown, the specific execution process of the present invention includes:

[0017] 1. Obtain historical interaction data between users;

[0018] 2. Construct the interaction probability matrix ω between users through historical data between users ij ;

[0019] 3. Apply the iterative formula to the entire network and iteratively calculate the user's influence I and susceptibility S;

[0020] 4. Label the top 1% of highly susceptible users;

[0021] 5. When a public opinion event occurs, if highly susceptible users are close to the public opinion information, predictions are made for this type of users, and the further spread of public opinion is then predicted.

[0022] After studying the experimental data, it was found that under the social celebrity communication model, if highly susceptible users are predicted, the ability to predict the spread of public opinion is stronger than that of highly influential users. That is, if highly susceptible users are predicted, the spread of public opinion can be effectively predicted. Here, social celebrities refer to users who are in a central position in the social platform and have a lot of attention. Based on the above empirical research, the present invention proposes a strategy that can effectively predict the spread of public opinion. When a public opinion event occurs on the network platform, and such public opinion events cannot block or shield social celebrities, the social networking platform can predict highly susceptible users on the platform, and then predict the further fermentation of the public opinion event.

[0023] After studying the experimental data, it was found that under the celebrity communication model of social platforms, highly susceptible users have a significantly stronger ability to influence the spread of public opinion than highly influential users, which means that they can effectively influence the spread of public opinion. Here, social celebrities refer to users who are in a central position in the social platform and have a lot of attention. Based on the above empirical research, we proposed a strategy that can effectively detect the spread of public opinion. First, the user's susceptibility is calculated through the user's historical interaction data, and highly susceptible users are marked. When a public opinion event occurs on the network platform, the social network platform can detect the highly susceptible users associated with the public opinion event, thereby controlling the further fermentation of the public opinion event.

Claims

1. A method for predicting the spread of public opinion. It is characterized in that The following steps are involved: The users in the social network are defined by attributes, and the attributes influence I and susceptibility S are given respectively, where influence refers to the ability of a user to influence other users, and susceptibility refers to the degree of influence by other users; thus, the probability p of user i's behavior affecting user j is ij The correlation is the influence of user i and the susceptibility of user j, denoted as p ij =I i S j ; Collect historical interaction data of users within a set time period through social platforms and estimate the interaction probability between users ω ij ; Iterative calculation of user influence and susceptibility: Among them, f i =∑ j A ij ω ij ,g j =∑ i A ij ω ij ,A ij is an element in the adjacency matrix A. When A ij =1, indicating that user j has interacted with the information posted by user i at least once, otherwise user j has not interacted with the information posted by user i; when iterates to When the value of is less than the set threshold, the iteration stops, where N is the number of users in the network and t is the number of algorithm iterations; Through iteration, the influence and susceptibility of all users in the social network are obtained, and the users ranked in the top 1% in susceptibility are marked. When online public opinion spreads, the marked users are monitored and the information they post on the social network is obtained to predict and monitor the spread of public opinion.

Images