Identifying the Offensive Words in the Twitter Using Latent Semantic Analysis

Abstract

The technologies around social media has changed how people connect with others, how they access information even how they organize their political point of view. The vast technologies has made the message can be sent quickly, become widespread and even viral. The drawback of this technology is people can post anything, from the positive content to negative content and it can be viral. Information contains negative content or known as hate speech can encourage conflicts between groups in society. In Indonesia, there is no automated mechanism to detect whether the information in social media is hate speech or not. Therefore, it could takes time to identify a hate speech. A starting point to identify hate speech is the use of offensive words and slurs the in feature. Once the offensive words are identified, we can classified the tweet into hate speech or not. Machine learning can be used to identify potential hate speech in collection of text. This paper aims to identify the offensive words using machine learning approach.