Sentiment Analysis of Mixed-Case Language using Natural Language Processing

Social Networking has become a platform for people to post their views and opinions about everything. People tend to mix their regional language with other languages that they are familiar with which gives rise to Mixed-Case Languages. These multilingual texts pose a great challenge to conventional Natural Language Processing systems, which currently rely greatly on monolingual resources and cannot handle the combination of multiple languages. The study of people's attitude, opinions, views and to classify them as positive, negative or neutral is called Sentiment Analysis. It is an extremely powerful tool in today's world since it allows businesses, restaurants, internet providers, airlines, etc to quickly detect dissatisfied customers, categorize issues with urgency and prioritize responses. The proposed system focuses on the sentiment analysis of a Hinglish (Hindi-English) text corpus. This paper discusses various approaches such as the Lexicon-Based, Rule-Based and Machine Learning approaches to study the effectiveness of the approach chosen for classifying the text corpus with their appropriate sentiment labels. The proposed Machine Learning techniques are evaluated using parameters Precision, Recall, F1-score and Accuracy and out of all the approaches used, the Support Vector Machine (SVM) and Logistic Regression (LR) approaches have given the best results with both algorithms giving an F1-score of 0.86 and accuracy of 86%.