Font Size: a A A

Psychological Factors Prediction And Application Of Social Network Users

Posted on:2016-05-21Degree:MasterType:Thesis
Country:ChinaCandidate:L ZhangFull Text:PDF
GTID:2308330464469109Subject:Software engineering
Abstract/Summary:PDF Full Text Request
The explicit behaviors of human highly correlates with our implicit psychological factors. Restricted by traditional methods of psychology, people have limited knowledge about psychological factors and can’t get access to their psychological factors easily. In the virtual world, behaviors of human also correlates with psychological factors. Moreover, reachable and computable online behavior information makes it possible to analyze psychological factors in large-scale. With the growing of social network users, researches about psychological factors prediction of social network users will provide psychological theory with unprecedented fuel for popularization and application, which is significant.The main purpose of this study is to analyze and predict psychological indexes of social network(mainly Sina micro-blog) users, including finding the relationship between social network characteristics and psychological indexes, and optimizing the prediction process. This thesis proposed two feature extraction methods for Sina micro-blog texs, which make the prediction more accurate. The first feature extraction method is topic feature extraction, it’s the first time this method draw into this field. Another related research is about the setting of numbers of topic models. The second feature extraction method is called dynamic lexicon feature extraction, it’s an optimization of traditional lexicon method, including a lexicon expansion algorithm and a lexicon feature extraction program. The lexicon expansion algorithm can automatically expand psychological lexicons according to different corpuses, so that lexicon features can constantly adapt to different Internet language environments. What’s associated with dynamic lexicon method is a suicide lexicon, which is a refinement of general purposed psychological lexicons, it’s designed especially for suicide detection.From a scientific perspective, both of the two feature extraction methods described in this paper update the record of psychological indexes prediction. Other related methods enriches research tools in this field. From an engineering perspective, the proposed feature extraction programs, psychological indexes assessment and prediction program and the suicide monitoring system can work together and achieve a set of functions including prediction, evaluation and exhibition.
Keywords/Search Tags:psychological indexes prediction, topic model, social network analysis
PDF Full Text Request
Related items