A simultaneous topic and sentiment classification of tweets

Abstract

Social networks have been an emerging technology for communication among billions of users. One of the most popular social networks is Twitter. The popularity of Twitter comes from its simplicity since it allows users to exchange messages of short length that does not exceed 140 characters and takes the form of tweets. In this paper, we propose a model for performing a classification of tweets posted by the Twitter user based on a mixture of the topic and sentiment of those tweets. The proposed approach is new in that it creates a model that combines the processes of topic and sentiment classification of tweets simultaneously. Therefore, with this model, one can categorize tweets according to their topics and simultaneously assign them into different sentiments categories. The topic of the tweets in the basic experiment of the proposed approach is classified into five main different categories including: “political”, “commercials”, “educational”, “religious”, and “sportive”. Meanwhile, the sentiment of those tweets is classified into three main different categories including “positive”, “negative”, “neutral”. The effectiveness of the proposed approach is demonstrated on a real dataset that consists of various extracted tweets with different categories of topics and opinions. The empirical results show that our approach is very powerful in categorizing tweets according to topics and simultaneously assigning them into different sentiments categories.