PREDICTING PERSONALITY USING RECURRENT NEURAL NETWORK BASED ON MBTI TYPE INDICATOR FROM TWITTER

Abstract

Personality is a basic of how a human behaves, thinks, and makes a decision. These days young people are going to experts such as psychologists and psychiatrists to take a personality test to figure out their personality. But the problem with this is that young people have to spend a high cost to do the test and consultation. Besides, the testrequiredmany questions to be answered which provided by the expert. The test itself could be less accurate if one miss interpreted the questions. There is another way to avoid this thing. Social media has been a popular platform for young people to express their life. Posts and texts from social media could be obtained and analyzed. In this final project, a web application proposed to predict a person's personality based on tweets posts by Twitter users. The application will show the predicted personality type and show valuable information such as matches jobs, strengths, weaknesses, romantic relationship, and friendship for the person with such personality type.The language used is in English. A Recurrent Neural network is used to recognize and analyze sequential data like words and sentences. Long Short-Term Memory architecture is also used as an extension of Recurrent Neural Network that extends the memory. With this tool, young people can predict their and other people's personalities. This tool is also useful for companies to predict their applicant personality.