Hybrid Model for Sentiment Analysis of Whatsapp Data

Abstract

Sentiment analysis, also called opinion mining (OM), is an approach to analyzing the sentiments expressed in data. This approach uses Natural Language Processing to classify the data based on emotions in different classes. Various Social media data are analyzed to determine sentiment, and Machine Learning (ML) techniques classify the data. This study utilizes ML models to analyze sentiment in WhatsApp data. The sentiment analysis process includes some steps, such as to pre-process the data, extract the features, and classify the data. The initial stage contributes to the clean-up of raw data and transforms it to make it suitable for analysis. Feature extraction is a stage to retrieve a relevant feature from the pre-processed data that contribute to determining sentiment. Finally, machine learning algorithms classify data to determine the sentiments expressed in the text. This work proposes a voting classifier which is hybrid architecture comprising SVM, KNN, and a Decision tree. Python is executed to simulate the suggested algorithm, and its performance is evaluated based on accuracy, precision, and recall metrics. These parameters are useful in measuring the efficiency of the algorithm in accurately classifying the sentiments existing in the data.