Hybrid Sentiment Analysis: A Novel Integration of Corpus-Driven and Machine Learning approaches to perform Sentiment Analysis of Kannada Political Tweets

Abstract

In today's digital age, understanding user opinions holds significant value in gauging product satisfaction and shaping consumer choices. Sentiment Analysis, a subset of opinion mining, focuses on evaluating emotions and viewpoints expressed towards specific subjects. This research dives into the realm of Kannada sentiment analysis, a regional language spoken predominantly in India. The challenge lies in extracting insights from limited Kannada corpora, acquired from https://gadgetloka.com using Python-Beautiful Soup and an API. Manually sifting through vast, unstructured data is a daunting task, prompting the application of an automated system called 'Sentiment Analysis or Opinion Mining.' This system adeptly dissects and extracts user insights from reviews, categorizing them into Positive, Negative, or Neutral sentiments based on their assigned weights. The research involves the analysis of Kannada tweets collected from various political entities and politicians to classify them as Positive, Negative, or Neutral interpretations. This classification holds the potential to assist political parties and politicians in refining their image and credibility. The study leverages Support Vector Machines, a supervised learning algorithm, to optimize the classification process through a set of hyperplanes, yielding enhanced accuracy.