An Integrated Approach for Analysing Sentiments on Social Media

Abstract

Sentiment analysis is an analytical subfield of Natural Language Processing (NLP) to determine opinion or emotion associated with the body of the text. The requirement for social media sentiment analysis has exceptionally increased with the growing extent of online activities in form of user generated content like posts or comments on social networking platforms. People often share their thoughts, opinions and reviews openly which can further be leveraged to analyze what they feel about a particular topic or their reviews/ feedback about a certain service. This study covers different approaches to conduct social media sentiment analysis on Twitter dataset both balanced and imbalanced obtained from Kaggle. For text analysis, we have implemented various classification techniques such as: Naive Bayes Classification and Support Vector Classification (SVC). It was concluded that SVC on twitter dataset surpassed other classification techniques in terms of performance.