Word2Vec-based Personal Trait Computing From User-generated Text

Personal trait is a habitual pattern for measuring behavior,thoughts,and emotions.It varies over individuals and is relatively stable in different situations over time.The personal trait is of great significance since it can be used into many applications,such as recommendation system,chatbot,and human resource management.In recent years,with the development of social media and wearable devices,more and more personal data are obtained.Many scholars study how to analyze and make good use of these personal data.Most researchers use supervised learning method to calculate the user profile,behavior characteristics and personality of users with labeled data.Most of the existing studies focus on user profile,behavior and personality.On the one hand,because the user profile and behavior characteristics are the manifestation of personal characteristics,we can not accurately describe the internal characteristics of a person through the user profile and behavior characteristics.The calculation of personality is usually measured by the big five personalities,which is obscure to non psychologists.Therefore,how to obtain specific and easy to understand personal trait becomes very important.On the other hand,there are some problems such as the high cost and the insufficient guarantee of the accuracy in obtaining the accurately labeled personal data.Therefore,there are some limitations in using supervised learning method to calculate personal trait.Therefore,this paper proposes a general method based on Word2 vec to calculate the specific and easy to understand personal trait.The calculation process mainly includes three aspects: topic word extraction,personal trait matrix generation and personal trait calculation.First,we use a variety of possible methods to extract a specific aspect of the topic words from the user generated-text,then use word2 vec method to transform the topic words into word vectors,and then put the word vectors into the personal feature matrix to analyze in the vector space model to obtain the personal characteristics.In addition,this paper also uses its proposed method to conduct a case study to verify the effectiveness of the proposed method.Finally,the results of this paper are verified indirectly and analyzed and discussed.