Hybrid Transformer–Recurrent Modeling for Sentiment Analysis in Low Resource Language

Read full text Summarise with VIRAT
Authors: Sheezan Farooq, Rumaan Bashir
Cited by: 0
References: 17

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

Sentiment analysis in low-resource languages like Kashmiri is underexplored due to the lack of annotated datasets and computational tools.This research proposes an effective hybrid deep learning model that combines the strengths of XLM-RoBERTa and BiLSTM networks for sentiment analysis of Kashmiri language text.The Kashmiri language, being lowresource and morphologically rich, poses significant challenges for natural language understanding tasks.A major contribution of this work is the creation of a manually annotated sentiment dataset tailored for the Kashmiri language, encompassing positive, negative, and neutral sentiment categories.This dataset serves as a foundational resource for training and evaluating sentiment classification models in this underrepresented language.The hybrid model combines XLM-RoBERTa transformer contextual embeddings with BiLSTM modeling sequences because of its ability to function in lowresource environments.Experimental results demonstrate that the hybrid model achieves state-of-the-art performance with a validation accuracy of 94.7% and an F1-score of 0.94 across all sentiment classes.Additionally, ROC analysis confirms high discriminative ability with an AUC of 0.99 for each class.Experimental findings demonstrate that integrating pre-trained transformers with recurrent models improves sentiment recognition abilities in Kashmiri by a significant degree.

Fields of Study & Subjects

Sentiment Analysis and Opinion Mining Topic Modeling Hate Speech and Cyberbullying Detection

Publication details

Journal / source
International Journal of Computer Applications
Volume
187
Issue
86
Pages
14–18
Publication date
20 March 2026
Language
EN