International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 04 | Apr 2025
p-ISSN: 2395-0072
www.irjet.net
Uncovering Emotions and Opinions of the Customers in WhatsApp Chats Conversations via Machine Learning Approach Neetesh kumar Nema1, Dr. Vivek Shukla2, Dr. S R. Tandan3 1Research Scholar, Department of Computer Science and Engineering,Dr. C.V. Raman University, Bilaspur, CG,India 2Assistant Professor, Department of Computer Science and Engineering,Dr. C.V. Raman University, Bilaspur,CG 3Assistant Professor, Department of Computer Science, Govt. R. V.R.S.G.C.K, Kabirdham, CG, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – Analyzing sentiment in WhatsApp chat data has
Sentiment analysis has evolved significantly—from early rule-based systems to more advanced machine learning, and more recently, deep learning techniques. The earliest approaches relied on sentiment lexicons and predefined dictionaries for classification (Turney, 2002). With the rise of machine learning, supervised algorithms like Naïve Bayes and Support Vector Machines (SVM) became popular, though they often struggled to interpret the informal and conversational nature of chat messages (Pang & Lee, 2008). In contrast, modern deep learning models—such as Recurrent Neural Networks (RNNs) and transformer architectures—have shown improved ability to understand context and subtleties in dialogue (Vaswani et al., 2017).
The study analyzed a dataset of WhatsApp messages, sorting them into positive, negative, and neutral sentiment categories. The results demonstrated that machine learning techniques provided more accurate sentiment classification than traditional rule-based approaches. Among all models tested, deep learning methods—particularly those based on bidirectional Long Short-Term Memory (Bi-LSTM) networks— achieved the highest accuracy. The discussion also emphasizes the inherent challenges in analyzing sentiment within informal conversations, especially due to factors like sarcasm and ambiguous context.
Recent studies have successfully applied sentiment analysis across a range of domains, including social media analysis (Liu, 2012), customer feedback interpretation (Cambria et al., 2017), and mental health assessment (Coppersmith et al., 2018). Nonetheless, analyzing informal chat data like that from WhatsApp remains challenging due to frequent use of slang, emojis, abbreviations, and sarcasm. Additional concerns involve maintaining user privacy, ensuring ethical handling of personal messages, and dealing with multilingual content.
become increasingly important for interpreting the emotional undertones of digital communication. This research examines how sentiment analysis can be leveraged to identify emotional patterns and communication styles within WhatsApp conversations. It begins by highlighting the significance of Natural Language Processing (NLP) in analyzing informal, unstructured text like chat logs. The study also reviews various analytical approaches, ranging from traditional machine learning models such as Naïve Bayes and Support Vector Machines (SVM) to advanced deep learning techniques like Recurrent Neural Networks (RNNs) and transformer-based models.
This study seeks to tailor and apply sentiment analysis methods specifically to WhatsApp chat data, aiming to not only capture emotional expressions but also to detect patterns in communication behavior, mood changes, and interpersonal interactions. By enhancing sentiment classification accuracy in informal settings and addressing privacy-related concerns, the research aspires to create a more nuanced and ethically responsible sentiment analysis model. Such a model could have valuable applications in fields including marketing, behavioral science, and social research.
Keywords: WhatsApp chat, Image, Sentiment Analysis, Emoji Analysis, NLP, Feature Engineering.
1. INTRODUCTION Sentiment analysis, often referred to as opinion mining, involves the computational examination of emotions, opinions, and attitudes conveyed through text. In recent decades, it has emerged as a key area within natural language processing (NLP), especially for interpreting content from digital communication platforms. With more than two billion active users worldwide (Statista, 2024), WhatsApp serves as a valuable source of conversational data, capturing a wide range of emotional and social interactions. As WhatsApp continues to be a primary medium for personal, educational, and professional communication, analyzing its chat data through sentiment analysis offers meaningful insights into user emotions, social dynamics, and even consumer behavior trends.
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 1202