Sentimental analysis and machine learning: hybrid methods and future prospects

Abstract

Sentiment Analysis (SA) using natural language is one of the most exciting topics of this decade due to the widespread use of social media, which allows for the analysis of interactions between individuals and the determination of important structures in these communications. SA is defined as a field of study that utilizes intelligence methods to analyze, process, and reveal the feelings, emotions, and sentiments of individuals hidden behind text or interaction through natural language processing. SA employs machine learning (ML) techniques to draw necessary inferences from user interactions. ML algorithms such as unsupervised, supervised, semi-supervised learning, and deep learning techniques are used to extract useful information from sentiments. Collecting individuals' opinions and making decisions based on them can be beneficial for many people, and through this, useful business insights can be gleaned. However, handling massive and multilingual data, determining appropriate sentiment polarity, dealing with sarcasm and emojis, and selecting the appropriate ML technique to build the analysis model are some of the challenges faced by SA. This chapter provides a detailed overview of trends and challenges in using ML in SA. To establish future objectives, this chapter examines the difficulties associated with SA. This article uses a state-of-the-art methodological approach. It provides a detailed overview of the trends, challenges and different methods of sentimental analysis, while incorporating elements of theoretical research. This methodology makes it possible to synthesise existing knowledge on the subject and explore future research directions.

Keywords: Machine Learning; Natural Language Processing; Opinion, Polarity, Sentiment Analysis.

Classification JEL: G41.

Paper type: Theoretical Research.