Personality Classification on Social Media Using Lexicon Construction with Decision Tree Algorithm

Sari

The development of information and communication technology has led to the emergence of various social media platforms that allow individuals to interact and express themselves. One important aspect that can be analyzed from data on social media is the user's personality. This research aims to classify the personality of social media users by using a lexicon-based approach optimized through the Decision Tree algorithm. The Decision Tree algorithm was chosen for its ability to partition data into smaller subsets based on specific features, resulting in an effective predictive model. This research uses data from the Kaggle platform and focuses on text analysis to identify personalities based on 16 personality types defined by the Myers-Briggs Type Indicator (MBTI). The classification process involves several stages, including data preprocessing, TF-IDF weighting, and application of the Decision Tree model. The results show that the lexicon-based approach implemented with the Decision Tree algorithm can produce high accuracy in social media user personality classification. This research contributes to the development of personality classification models that can be used in various applications such as marketing, workforce recruitment, and product development. However, this research also recognizes the limitations in the accuracy of the model which is influenced by the variability of language use in social media.