Gravity based Punjabi Question Answering System

Abstract

Question and answers systems are foundation for building intelligent information retrieval systems but, current server of these show that limited effort has been in building Punjabi Question Answer systems. This paper is an attempt to overcome this limitation. The research work is based on a concept taken from physics: Point of Gravity. In this approach the question and answer text are processed to extract numerical features so as to determine Point of Gravity. Matching Gravity Score values are computed for finding answer against a question query. Series of random evaluation sample sets show a high degree of overall system accuracy (above 91%) for each question type in terms of precision and recall. Individually the evaluation of each question type also shows not less than 91% accuracy in terms of precision.

Particle Swarm Optimization for Punjabi Text Summarization

“…(Kaur and Gupta, 2014) have discussed the concept of sentiment analysis in the Punjabi while using lexicon, unigram and simple scoring method. (Singh and Kaur, 2014) have presented a simple and easy to implement rule-based approach for shallow question generation system using NER tool. The tool has recognized names from the Punjabi text containing historical information, and generated appropriate questions from them.…”