Main Concepts, State of the Art and Future Research Questions in Sentiment Analysis

It is argued that there is no hard evidence that fuzzy sets or hybrid approaches encompassing unsupervised learning, fuzzy sets and a solid psychological background of emotions could not be at least as effective as supervised learning techniques.

Abstract

This article has multiple objectives. First of all, the fundamental concepts and challenges of the research field known as Sentiment Analysis (SA) are presented. Secondly, a summary of a chronological account of the research performed in SA is provided as well as some bibliometric indicators that shed some light on the most frequently used techniques for addressing the central aspects of SA. The geographical locations of where the research took place are also given. In closing, it is argued that there is no hard evidence that fuzzy sets or hybrid approaches encompassing unsupervised learning, fuzzy sets and a solid psychological background of emotions could not be at least as effective as supervised learning techniques.