Personality Prediction Model Based On User Sentiment Analysis And Network Relationship Analysis

With the development of the Internet, various Internet applications become more and more important in people’s lives. People’s habits and demand are different when using the internet. Therefore, personalization has gradually become an important factor in the applicat ion of Internet recommendation system and advertising display. People’s behavior and preferences are often associated with the personality of the individual. The personality of network users can promote the development of personalized applications. The traditional personality measurement is mainly relies on the psychological researchers to interview the research object or to let the research object to fill in the personality questionnaire to obtain their personality data. These methods all require a lot of manual operation. It is difficult to carry out large-scale user personality measurement. It is not suitable for the Internet environment. The behavior of people on the internet is closely related to personal personality. In recent years, social networks develop very fast. Social networks users will produce a lot of information and behavior data. It will be high-efficient to predict the personality using social network data. In this paper we study the Sina Weibo, extract the user’s characteristics and personality characteristics, and build the prediction model. The experiment verifies the feasibility of the model. The work of this paper content can be summarized as follows:(1) The existing researches of personality prediction usually only consider the user’s statistical characteristics. This paper proposes a new method to predict the user’s personality. The sentiment characteristics are extracted through the sentiment analys is. This paper also extracts the user’s network relationship characteristics and the inherent characteristics, then build the user personality prediction model. The automatic prediction of user ’s personality through the microblog data is realized.(2) The microblog text contains a large number of network language and facial expression symbol. So the traditional sentiment dictionary is not suitable for sentiment analysis of microblog. This paper expands the traditional sent iment dictionary combined with the microblog network language and builds a microblog expression symbol dictionary. At the same time, the negative dictionary and double negative dictionary are added into traditional sentiment dictionary. Sentiment analysis of the microblog text is conducted with the improved sentiment dictionary.(3) The existing researches of user personality prediction are usually only consider the user’s attributes. This paper studies the characteristics of the user’s network relationship, which mainly considers the similarity and interaction between users and friends.(4) In this paper, we build the continuous prediction model and the classification predication model based on the different forecast demand. Based on the analysis of the different regression methods, a multi task learning method is put forward to establish a continuous prediction model. A support vector machine algorithm is used to establish the classification prediction model.