Sentiment Analysis of Tweets Using Emojis and Texts
Mohammed Saleh Alzubaidi, Farid Bourennani · 2021
Nowadays, social media generate massive volumes of data containing valuable information to several applications such as marketing, business, and politics which are interested to analyze opinions and sentiments of individuals. A lot of efforts have been directed towards sentiment analysis (SA) based on texts; however, limited efforts have been directed towards the study and analysis of other DTs such as emojis and other multimedia. The processing of other data types (DTs) not only can generate complementary information, but we argue that the combined processing of heterogeneous DTs (HDTs), called heterogeneous data mining (HDM), should lead to more accurate results. In this work, we propose the HDM of tweets using both texts and emojis for bi-sense SA; the SA results using HDM were more accurate than the exclusive use of either texts or emojis, and interestingly the use of emojis achieved a higher accuracy than texts.