Using K-Nearest Neighbor Algorithm for Personality Classification of Twitter’s Users Based on the Big Five Theory

Abstract

Social media is an application or website-based system that enables users to create and share content or participate in social networking that allows its user to share their thoughts, opinions, or feelings that represent their personality. At present several studies to classify an individual's personality through social media have been developed, especially on social media Twitter. However, most of the analysis on Twitter only uses text based data such as posted tweets. This research presents a study on analyzing the users’ twitter data to classify their types of personality based on Big Five Theory by using their social statistic data. The data were acquired using Twitter API which was taken from Indonesian users with the total of 225 data. This study shows that using K-Nearest Neighbor (K-NN) Algorithm for classification of these data were not resulting in high accuracy. However, this study has shown that amount and balance distribution of training data critically contribute to the performance of classification process.