Multilingual Customer Feedback Analyzer Using MuRIL
DOI:
https://doi.org/10.7492/y6rwhq79Keywords:
Multilingual NLP, MuRIL, Customer Feedback Classification, Indic Languages, Transformer Models, Cross-Lingual Learning, Sentiment Analysis, AI-driven AnalyticsAbstract
Customer feedback is considered one of the richest resources for an organization. It provides value regarding knowledge on product
performance, user satisfaction, and areas that need improvement. However, continuous analysis of a large volume of unstructured feedback in
diverse languages is very time consuming and inefficient. This paper presents a multilingual approach to automatically classify reviews in the
English, Hindi, and Gujarati languages into categories like bug reports, feature requests, positive feedback, and others. The proposed system
uses MuRIL (Multilingual Representations for Indian Languages), a transformer-based model designed for Indic languages. This model
captures contextual and cross-lingual meanings. This model results in strong performance when fine-tuned on a balanced multilingual dataset
for classification across various language structures. This framework is scalable, automated, and cloud-based feedback analysis. It supports the
organizations in making quicker, data-driven choices, improving product quality, and increasing overall customer satisfaction.
