Question Answering System for Tamil Using Deep Learning

Abstract

Tamil, a Dravidian language family member, is widely spoken in numerous Indian states. But languages like Tamil, are under- represented on the web. Many NLP models perform worse with these languages when compared to English, the effects of which lead to subpar experiences in many web applications for most of the users. The number of Tamil websites has grown and continues to expand. Tamil newspapers, periodicals, and e-journals are examples. Therefore Tamil Question Answering System will be beneficial to many people who are in the rural areas and also for those who prefer our native Tamil Language. This paper aims to create a Question Answering System in Tamil using deep learning techniques, starting from question processing through an interface, retrieving the relevant context for the given question and finally the answer is displayed after appropriate processing. The model is evaluated using a manually created test dataset and the metrics are analyzed.