An overview of Multimodal Sentiment Analysis research: Opportunities and Difficulties

Abstract

The scatter form of multimedia data such as text, image, audio, and video posted regularly in the social media may contain useful information for the organizations. But, this information should be derived with the use of some form of analysis known as Multimodal Sentiment Analysis (MSA). But, there is a lack of proper analytic tools for such analysis. This paper presents a thorough overview of more than fifty most recent MSA research articles to find the gaps in terms of tasks, approaches theories and applications used till date. There seems to be no single approach, theory, and tool which can support MSA. The study showed that each and every mode presents different difficulties which have not bee n fully solved yet, such as feature points of a face, voice clarity in audio, video summarization and so on, and are great research opportunities for the future researchers. Also, this research recommends a list of existing and upcoming difficulties and opportunities of MSA research.