Survey and Improvisation for Sentiment Analysis of Textual Content

Abstract

Sentiment Analysis (SA) is useful to extract important information from textual content such as blogs, tweets, social media, websites, to name a few. Reviews derived from SA are useful for both the user and an organization. For example, if someone likes a product and want to purchase it, they will look at its reviews before making a purchase. Similarly, if an organization has launched a product, they'd need reviews from its users to improvise their product. SA extracts useful information from surveys, tweets, or social media content. This survey aims to give the readers an overall idea of how SA works. Figure. 1.1 describes the generalized ow of SA. This report presents SA procedure in detail including different levels, pre-processing techniques, types of features and feature selection techniques, approaches, challenges in SA, recent work in SA and underline results. Report also presents overview and improvement in recent approaches such as Fuzzy sentiment phrase approach, aspect based SA with auxiliary sentence construction and enhanced naive bayes approach with simulated and improved accuracy result.