SENTIPUBLIKO: Sentiment Analysis of Repost Jejemon Messages using Hybrid Approach Algorithm

Abstract

Jejemon language becomes a form of communication dialect. It was a form of expression used by a particular social group unknown as Jejemon. However, the Jejemon expression has different formats ranging from basic form of changing letter to number, lowercase letter to uppercase letter, inserting shortcut texts into more complicated format. This paper aims to classify Jejemon tweet whether it is a positive, negative or neutral sentiment through sentiment analysis techniques. Experiment included translation of Jejemon formatted tweet, reduction of sentiment scores on repost tweets and sentiment classification. Analysis of experiment results involves Paired T-Test, confusion matrix, precision, recall, f-score and accuracy. Evidently, translated Jejemon tweet resulted 78.5% similar from the actual message using cosine similarity algorithm. Furthermore, Paired T-Test shows no significant difference between new sentiment scores from translated expression and actual sentiment scores using Hybrid Algorithm. Sentiment analysis metrics such as precision, recall, f-score and accuracy show acceptable values of 71%, 76%, 71% and 73% respectively.