Improving the Performance of RDQA Using Lexical Based Inference Extraction
Published: 2014
Author(s) Name: Renita Raymond, Karnavel Kuppusamy |
Author(s) Affiliation: Computer Sc. and Engg. Department, Anand institute of Higher Technology, Chennai, Tamil Nadu, India.
Locked
Subscribed
Available for All
Abstract
This paper presents an enhanced approach for Question Classification and Answer Extraction in Restricted Domain Question Answering (RDQA). Question Classification and Answer Extraction is the core problem of RDQA and determines the performance of the Question Answering in the Restricted Domain. The proposed approach improves the performance of
RDQA by means of (1) Question type prediction model based on Bayesian classification (2) Lexicalized-Index based Passage Retrieval (3) Lexical-Semantic based Inference Extraction. This paper also describes user-centered task-based evaluations for Answer Validation. Further improvements are achieved by combining our model with the classic one to improve the performance of Restricted Domain Question Answering.
Keywords: Restricted Domain Question Answering (RDQA), Bayesian Classification, Passage Retrieval, Answer Extraction, Text Inference
View PDF