Machine Learning Techniques for Opinion Mining from Social Media

Author(s)

Abstract
Expressing emotions through various channels is part of human life. Directly or indirectly, we somehow reflect our opinions through speech, writings, etc., in our daily life. Opinions containing emotional or sentimental words have huge impact in the society. Analyzing the emotions and sentiments of people has its own importance. For example, we can measure the well being of a society, prevent suicides, and measure the degree of satisfaction of their customers by analyzing the comments or the feedback. The world wide web sites like social media, forums, review sites, and blogs generate a large volume of data in the form of opinion, emotion, and sentiment about social events, government policies, political events etc. Increased use of technology has made people proactively express their opinion through social media sites like Twitter, Facebook, and Instagram. Decision makers can make use of social media content to understand how people react to policies, events, and consumer products. But, social media analytics is a complex task due to the challenges in the natural language processing of social media language. These messages do not adhere to grammatical standards. The unstructured data from the social media needs to be cleansed and well-structured for opinion mining. These messages often reflect the opinion, emotion, and sentiment of the public through a mixture of text, emoticons, image, etc. Standard natural language processing tools cannot be used to analyze the emotion or sentiment hidden in the social media content. Social media users use emoticons like smiling face (), angry face () etc., to express emotion instead of words. They also express positive () or negative () sentiment using emoticons. These statements are called electronic Word of Mouth (eWOM) and are much popular in business and service industry to enable customers to express their point of view. We propose to use a two-step approach for opinion mining of social media content. Instead of using language parsers for parsing the eWOM, we propose to use machine learning algorithm for opinion mining.

Keywords
Emotion Analysis, Sentiment Analysis, Opinion Mining, Social Media Analytics.

Cite this paper
K. Victor Rajan, Freddy Frejus, Machine Learning Techniques for Opinion Mining from Social Media , SCIREA Journal of Computer. Volume 7, Issue 1, February 2022 | PP. 10-29. 10.54647/computer52271

References