Enhancing Multilingual Sentiment Analysis with Explainability for Sinhala, English, and Code-Mixed Content

2025-04-18

AI Summary

Low-resource languages (e.g., Sinhala) and English–Sinhala code-mixed texts exhibit poor performance and limited interpretability in multilingual sentiment analysis for banking applications. Method: We propose the first explainable, fine-grained multilingual sentiment analysis framework tailored for financial reputation management. It innovatively integrates domain-specific lexicon enhancement, heterogeneous pre-trained model fine-tuning—XLM-RoBERTa for Sinhala and code-mixed texts, BERT-base-uncased for English—and real-time, visualizable fine-grained attribution via SHAP and LIME. Contribution/Results: Experiments yield 92.3% accuracy (F1 = 0.89) on English, and 88.4% on Sinhala and code-mixed data. The system is deployed as a user-friendly business interface, enabling operational fine-grained sentiment monitoring and auditable decision traceability in financial contexts.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Sentiment analysis is crucial for brand reputation management in the banking sector, where customer feedback spans English, Sinhala, Singlish, and code-mixed text. Existing models struggle with low-resource languages like Sinhala and lack interpretability for practical use. This research develops a hybrid aspect-based sentiment analysis framework that enhances multilingual capabilities with explainable outputs. Using cleaned banking customer reviews, we fine-tune XLM-RoBERTa for Sinhala and code-mixed text, integrate domain-specific lexicon correction, and employ BERT-base-uncased for English. The system classifies sentiment (positive, neutral, negative) with confidence scores, while SHAP and LIME improve interpretability by providing real-time sentiment explanations. Experimental results show that our approaches outperform traditional transformer-based classifiers, achieving 92.3 percent accuracy and an F1-score of 0.89 in English and 88.4 percent in Sinhala and code-mixed content. An explainability analysis reveals key sentiment drivers, improving trust and transparency. A user-friendly interface delivers aspect-wise sentiment insights, ensuring accessibility for businesses. This research contributes to robust, transparent sentiment analysis for financial applications by bridging gaps in multilingual, low-resource NLP and explainability.

Problem

• Improves sentiment analysis for low-resource languages like Sinhala
• Enhances explainability in multilingual sentiment classification
• Addresses accuracy gaps in code-mixed and banking feedback analysis

Innovation

• Hybrid aspect-based sentiment analysis framework
• Fine-tuned XLM-RoBERTa for multilingual text
• SHAP and LIME for interpretable outputs