InstaSent: A novel framework for sentiment analysis based on instagram selfies

Abstract

Sentiment analysis and opinion mining is the field of study to analyse opinions, attitudes, and emotions of people. It is the most studied research field. With the evolution of social media, many new terms have been evolved including selfies. A selfie has provided a way for to the users of social media to record their personal memories. People are sharing their images on social media in a massive amount. Keeping in mind the big data nature, it is very difficult to analyse selfies manually. In this research study, we have proposed a framework called InstaSent for sentiment analysis based on Instagram selfies. This framework incorporates both text mining and image mining techniques for sentiment prediction. Support vector machine is used for sentiment classification based on the text associated with selfies like captions, hashtags, comments, and emoticons, while the deep learning method Convolutional neural network is used for processing image data for sentiment analyses. By combining the practices for text mining and image mining we believe that this technique will outperform all other techniques presented in this domain. In a nutshell, we believe that our research study has provided novel opportunities for researchers to explore the use of selfies in other domains.