An Automatic Template Based Question Answering System to Improve E-Learning

Vol.2 No.4

Year: 2008

Issue: April-June

Title: An Automatic Template Based Question Answering System to Improve E-Learning   

Author Name: Ashish Gupta, Ankush Mittal 

Synopsis:   

This paper presents a fully automatic closed domain question answering system designed specifically to improve student's learning experience through E-Learning. The question answering system allows students to access course data available on distance education web sites in a very effective manner by allowing them to ask questions in natural language. It uses various natural language processing tools to give students relevant and quick answers. The question answering system has been designed to fully utilize the domain knowledge specific to a course to improve accuracy and speed of the system. It also utilizes domain knowledge to disambiguate ambiguous terms used in question. The system has been designed considering the requirements of students and can handle variety of questions types generally asked by students. This is achieved by implementing template based approach of answer extraction.  Many templates for factual type question have been implemented. The system can be targeted to any domain or subject by providing the domain knowledge in the form of domain keywords at initial setup. Word sense disambiguation algorithm is evaluated on Semcor corpus and has shown significant performance. Results for question answering system have been taken on course material of operating system with various types question including FAQ’s, expert and naïve questions.