International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 03 | Mar 2025
p-ISSN: 2395-0072
www.irjet.net
Exploring Sentiment in WhatsApp Conversations: A Machine Learning Approach Pratishtha Dwivedi1, M Nikhileshwari2, Aviral Singh3, Priyanka Rajak4 1,2,3B.Tech Student Department of Computer Science and Engineering, LCIT Bilaspur, CG, India 4Assistant Professor, Department of Computer Science and Engineering, LCIT Bilaspur, CG, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Sentiment analysis of WhatsApp chat data is
becoming increasingly significant in understanding the emotional tone of digital conversations. This study explores the application of sentiment analysis to WhatsApp chat messages to uncover patterns of communication and emotional contexts within social interactions. The research begins by outlining the importance of natural language processing (NLP) techniques in sentiment analysis and emphasizing their utility in analyzing unstructured text data such as chat logs. The introduction details the various methods used, including machine learning models such as Naïve Bayes and Support Vector Machines (SVM), as well as deep learning approaches such as Recurrent Neural Networks (RNNs) and transformers. The study utilized a dataset of WhatsApp conversations, categorizing messages into positive, negative, and neutral sentiments. Results indicate that machine learning-based models outperform traditional rule-based methods in terms of accuracy, with the highest success rates observed in deep learning models, particularly those employing bidirectional LSTM (Long Short-Term Memory) networks. The discussion highlights the challenges of contextual ambiguity and sarcasm in interpreting sentiments, especially in informal communication settings, such as WhatsApp. Fig – 1: Block Diagram of SentimentAnalysis
Keywords: WhatsApp chat files, visualization, Sentiment Analysis, Emoji Analysis, Natural Language Processing , Feature Engineering.
The evolution of sentiment analysis has progressed from rule-based systems to machine learning approaches, and more recently, to deep learning methodologies. Initial models relied heavily on predefined lexicons and dictionaries for sentiment classification (Turney ,2002). As machine learning gained traction, supervised algorithms such as Naïve Bayes and Support Vector Machines (SVM) became prevalent, although these models often encountered difficulties with the casual and colloquial language typical of chat-based interactions (Pang & Lee, 2008). In recent years, deep learning models, including Recurrent Neural Networks (RNNs) and transformers, have demonstrated greater potential in grasping context and subtleties in conversational text (Vaswani et al., 2017).
1.INTRODUCTION Sentiment analysis, also known as opinion mining, is a computational study of emotions, opinions, and attitudes expressed in a text. Over the last few decades, sentiment analysis has become a prominent field in natural language processing (NLP), particularly for analyzing data from digital communication platforms. WhatsApp, with over two billion active users globally (Statista, 2024), represents a rich source of conversational data, reflecting a variety of emotional and social exchanges. Given the increasing reliance on WhatsApp for personal, educational, and business communications, sentiment analysis of WhatsApp chat data is of great interest for understanding user sentiment, social behavior, and even market trends.
© 2025, IRJET
|
Impact Factor value: 8.315
Current research demonstrates successful implementations of sentiment analysis across various domains, including social media monitoring (Liu, 2012), consumer feedback analysis (Cambria et al., 2017), and mental health screening (Coppersmith et al., 2018). However, challenges persist in
|
ISO 9001:2008 Certified Journal
|
Page 266