Evaluating the Natural Language Understanding of a Machine by Answering Multiple Choice Questions for a Comprehension Text using Proposed LKD Model
K.M. Arivu Chelvan1, K. Lakshmi2
1K.M. Arivu Chelvan, Research Scholar, Department of CSE, Periyar Maniammai Institute of Science & Technology Thanjavur (Tamil Nadu), India.
2K. Lakshmi, Professor, Department of CSE, Periyar Maniammai Institute of Science & Technology Thanjavur (Tamil Nadu), India.
Manuscript received on 30 March 2019 | Revised Manuscript received on 09 April 2019 | Manuscript Published on 27 April 2019 | PP: 891-896 | Volume-7 Issue-6S2 April 2019 | Retrieval Number: F11060476S219/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Machine Reading (MR) is an art of understanding text by the machine and one of the best tools to evaluate the understanding level of the machine is Reading Comprehension System (RCS) with Multiple Choice Questions (MCQ). In this paper, we proposed with a new knowledge representation, for understanding the given text, called Linguistic Knowledge Document (LKD). Such, LKD is generated from the given comprehension text. Natural Logic is used for generating the LKD. It is like an inference engine which contains all possible inference for each sentence in the comprehension text. The proposed LKD is acting like a human brain for the machine for answering the questions inquired by MCQA system. We use token based alignment model for finding answers from the LKD. We evaluate our system on RACE dataset and the obtained results are compared with recent methods. The comparison results show that the proposed model outperforms the recent results.
Keywords: Machine Reading; Reading Comprehension System; Multiple Choice Questions; Natural Language Inference; Alignment; Natural Logic.
Scope of the Article: Machine Learning