Degree

Master of Science in Computer Science

Faculty / School

Faculty of Computer Sciences (FCS)

Department

Department of Computer Science

Date of Submission

2020-06-30

Advisor

Dr. Sajjad Haider, Professor, Faculty of Computer Science, Institute of Business Administration (IBA), Karachi

Project Type

MSCS Survey Report

Abstract

Commonsense reasoning is an active area of research in Natural Language Processing (NLP). Tasks and datasets involved in this problem are designed to ensure that the corresponding models go beyond pattern recognition and incorporate commonsense knowledge. This research survey presents five commonsense tasks: Winograd Schema Challenge, Winograd NLI (WNLI), Situations with Adversarial Generations (SWAG), CommonsenseQA (CQA), and Event2Mind. The survey examines nine papers, including models, experiments, results, and datasets. The survey also compares the models in terms of performance on the relevant tasks and recommends future direction. This work aims to provide readers a profound understanding of commonsense tasks in NLP and the relevant models which solve them.

Notes

Five commonsense tasks and their relevant models were presented in this survey. Although many commonsense tasks have been proposed in the literature, these five tasks were carefully selected to bring diversity. The purpose of commonsense reasoning tasks is to ensure that the corresponding models go beyond pattern recognition and incorporate commonsense knowledge. This survey presented nine papers with comprehensive coverage of their models, experiment, results, and datasets. It intended to provide readers with a profound understanding of commonsense tasks in NLP and the relevant models which solve them.

The full text of this document is only accessible to authorized users.

Share

COinS