Personality Traits Analysis of Instagram Social Media Community in Terms of Deep Learning Method

Abstract: Nowadays, social media has become the main medium for people to communicate with each other. People begin to care about the quality and effectiveness of making friends on social media. Knowing the personality of users can be used as a basis for judging new friends on social media. Therefore, how to predict the user’s personality through the activity records on social media has become an important issue. Due to the long length of traditional personality scale questionnaires, it takes a long time to judge the personality traits of users. In order to improve the efficiency of analyzing the personality traits of users, the purpose of this study is to provide a method for predicting personality traits using user pictures and posts, so as to reduce the time to judge the user’s personality. A total of 200 Instagram users’ pictures and posts were collected, using the RGB average value of the picture, the percentage of LIWC post vocabulary, and the combination of the RGB average value of the picture and the percentage of vocabulary in the LIWC post as features, and established CNN, SVM, Decision tree and Personality trait prediction models such as Logistic Regression are used to predict the five major personality traits and compare their accuracy. The results of this study confirmed that using the CNN algorithm combined with the image RGB average and the LIWC post vocabulary percentage as features has a high accuracy rate (85%) in predicting personality traits. In terms of predicting individual personality traits, CNN algorithm has a higher accuracy rate of 87.5% for rigorous personality, agreeable personality and neurotic personality; SVM algorithm has higher accuracy rate for extraverted personality and agreeable personality. Both are 87.5%; Decision is better at predicting rigorous personality, with an accuracy rate of 87.5%. This result can be used as a reference for future research.

Key words: Instagram, Big five personality traits, convolutional neural networks, social media, personality trait prediction