Research On User Emotion Analysis And Its Influence On Information Behavior In Information Security Field

With the acceleration of informatization process,the iteration speed of information supervision system is far less than the innovation of information technology,and information security incidents occur frequently.Faced with the loss of interests in information security incidents,users tend to vent their emotions through Internet means,especially social media platforms,which further worsens the network environment.According to the theory of social influence,users’ emotions will be affected by the development of the whole network environment and events,and different types of information security events will have different effects on users’ emotions,which cannot be generalized.Therefore,in order to maintain the "clear sky" of the network environment,it is necessary not only to understand the emotional changes of users in a period of time from a macro perspective,but also to understand the impact of emotional changes on users’ information behavior from a micro perspective,so as to regulate and soothe users’ emotions,control the fermentation and propagation speed of public opinion under different events,and reduce the loss of interests.Based on this,this study selects two types of events in the field of information security that are most closely related to users’ interests,namely,"hacker attack" event and "data leak" event.Firstly,the emotion classification of users at different stages of events under the two-dimension theory is analyzed by the emotion dictionary.Then,the BERT-LSTM model was selected to conduct emotion classification and effect comparison based on multi-label emotion labeling combined with automatic annotation and manual annotation.Finally,based on the multi-label emotion labeling corpus,this study draws on the emotional cognition theory and the two-dimension theory to compare and analyze the influence of social media emotions(emotional valence dimension and emotional arousal dimension)on information behavior(information adoption behavior and information interaction behavior)under these two types of events.It is expected to provide theoretical and practical basis for enterprises to sort out social media emotions,regulate information behavior,clean up network space,and safeguard personal and corporate interests.Based on the above research,the following conclusions can be drawn:(1)Based on the emotion analysis in the field of information security based on the emotion dictionary,it is preliminarily found that users are driven by interests and face information security events,their emotions fluctuate more under the arousal dimension;The emotional quantity of user text in the event of "data leak" is significantly higher than that in the event of "hacker attack",and users are more sensitive to the event of "data leak".(2)Under the same parameters,the classification effect of the Bert-LSTM model based on user emotion in the field of information security is better than that of BERT model and Bert-CNN model;(3)Negative emotion and low arousal emotion negatively affect information adoption behavior;Positive emotion positively affects information adoption behavior;Low arousal emotion positively affects information interaction behavior;Positive emotion and high arousal emotion negatively affect information interaction behavior;(4)The relationship between "data leakage" events and information behavior was affected by the negative moderating effect of positive emotion and the positive moderating effect of high arousal emotion;The relationship between "hacker attack" event and information interaction was positively moderated by negative emotion,and the relationship between information adoption behavior was negatively moderated by low arousal emotion.