Survey on sentiment analysis from affective multimodal content

Abstract

For sentiment analysis, online generated user content is important particularly in social media. People are using online media increasingly to say their opinions and share their knowledge’s through videos, tweets, and audio recordings. Analysis of such extensive content can help to extract user sentiments towards events or topic. With the advancement of social network for sharing reviews, recommendations, feedback, opinions, and ratings it has become a necessary for analysis. The emotions and sentiments helps in making critical decisions in organizations and businesses, and also situation awareness in environment for individuals. Since shared content on social media is multimodal in nature, research in affective computing has evolved over development of multimodal analysis frameworks. Natural language processing for text and emotion recognition from audio and visual is practice for analysis. This paper presents the overview of different techniques and approaches in sentiment analysis for text, audio, and visual modalities.