In Partial Fulfillment of the Requirements for the Degree of Master of Science

Huijie Li

will defend her thesis

Using Reasoning to Answer to Decision Questions Involving an Adjective

Abstract

Question Answering is an important topic under NLP. It would make NLP applications feasible to understand questions and give accurate answers. However, it is challenging to build a Question Answering system that can satisfy every user. In this thesis, we focus on solving the decision question including an opinion adjective. The method is developed at a knowledge based NLP system named A Learning Program System (ALPS). ALPS is an object-oriented NLP application. It includes basic NL analysis tools, like English grammar learning, parsing sentences, and storing and retrieving knowledge. We identify and address several sub-problems: the knowledge representation of opinion or notion, justification in the use of adjective for describing an opinion, logical inference through the related factors lists and learnt examples to give answers.

Human can easily learn notion. But, it is hard for computer to have the same learning ability. This thesis creates reasoning method to model procedure of answering notion or opinion decision question in program. The model firstly learns the notion, like importance. And then, it justifies the use of adjective for a notion. Finally, it will give answer to an opinion decision question through reasoning in factors list and examples list under this notion.

Date: Tuesday, November 29, 2016
Time: 2:30 PM
Place: PGH 516
Advisor: Kam-Hoi Cheng

Faculty, students, and the general public are invited.