Classifying User Personality Based on Media Social Posts Using Support Vector Machine Algorithm Based on DISC Approach

Abstract

Twitter is one of the largest social media with 326 million active users in January 2019. Indonesia emerged as one of the largest countries in terms of Twitter users. Every day more than millions of tweets are published by Twitter users. This study tries to analyze Tweets to get the personalities from chosen Twitter accounts by using the DISC character approach. The classification algorithm that will be used is Support Vector Machine (SVM) with Term Frequency-Inverse Document Frequency (TF-IDF) weighting on the dataset. This research starts with preprocessing stages such as Data Cleansing and Case Folding. We involved psychologists to validate the personality approach of 109 Twitter accounts to determine each Twitter user character. The character classification results used in this study are Dominance, Influence, Steadiness, Compliance (DISC). From 109 Twitter accounts, we considered as the final dataset, we obtain an accuracy of 36.37%, average precision of 23.11%, and average recall performance of 35.25%.