A RoBERTa-Driven Deep Feature Learning Framework for Real-Time Political Event Classification on Twitter

Read full text Summarise with VIRAT
Authors: S Swapna, Keesari Rajashekhar, Jallika Rakesh, Mohammad  Sadhik Pasha, Kogila Lavan Sai
Cited by: 0
References: 23

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Twitter has become a prominent platform for real-time information exchange, with over 500 million tweets generated daily, a substantial proportion of which are related to dynamic real-world events.Political incidents alone account for nearly 40% of trending discussions worldwide, making automated event monitoring essential for timely decision-making and crisis response.Manual classification of such high-volume textual data is labor-intensive, inconsistent, and impractical for large-scale or real-time analysis.Furthermore, conventional machine learning techniques often fail to capture contextual and semantic nuances necessary to distinguish diverse event categories accurately.To address these limitations, this study proposes an automated political event monitoring framework that integrates transformer-based embeddings with optimized classification strategies.The methodology begins with comprehensive Natural Language Processing (NLP) preprocessing, including tokenization, stop-word removal, normalization, and lemmatization, followed by Exploratory Data Analysis (EDA) to identify trends and data distributions.Context-aware semantic features are extracted using Lightweight Robustly Optimized Bidirectional Encoder Representations from Transformers (RoBERTa), enabling efficient yet rich textual representations.Unlike baseline classifiers such as standard Stochastic Gradient Descent (SGD), Histogram-based Gradient Boosting (HGB), Greedy Tree Classifier (GTC), and Random Forest Classifier (RFC), the proposed approach incorporates Deep Neural Network (DNN)-based feature selection combined with an optimized SGD classifier to enhance discriminative learning and handle class imbalance effectively.The system categorizes tweets into six event classes: disaster, political, positive, protest, riot, and terror.Experimental results demonstrate improved accuracy, scalability, and adaptability, making the framework suitable for real-time political event detection and supporting applications in governance, public safety, and crisis management.

Fields of Study & Subjects

Text and Document Classification Technologies Topic Modeling Sentiment Analysis and Opinion Mining

Publication details

Journal / source
Journal of Science Engineering Technology and Management Sciences
Volume
3
Issue
3
Pages
245–257
Publication date
1 March 2026
Language
EN