|
|
Type of Document Dissertation Author Toh, Huey Ling URN etd-03222010-133448 Title Design of a Low-cost Adaptive Question Answering System for Closed Domain Factoid Queries Degree Doctor of Philosophy Department Computer Science, Department of Advisory Committee
Advisor Name Title Lois Wright Hawkes Committee Chair Daniel Schwartz Committee Member R.C. Lacher Committee Member Theodore Baker Committee Member Ralph Dougherty University Representative Keywords
- Query-Based Indexing
- Adaptive Factoid Question Answering
- Closed Domain
- e-Learning
Date of Defense 2010-02-26 Availability unrestricted Abstract Closed domain question answering (QA) systems achieve precision and recall at the cost of complex language processing techniques to parse the answer corpus. We propose a query-based model for indexing answers in a closed domain factoid QA system. Further, we use a phrase term inference method for improving the ranking order of related questions. We posit that a query can be used as the unique identifier of an answer, and thus, the recognition of a query allows us to retrieve the correct answer. In instances where a query is unrecognized, we infer synonymous relationships with other queries through the use of a user feedback loop to improve the ranking order of closely related questions, where possible.
The goal of this research is to build a prototype as proof-of-concept that will learn domain-specific knowledge with increased usage through time. This study will focus its efforts in researching the feasibility of a lightweight QA learning system that adapts its responses based on the interaction amongst its users. This offers a lightweight approach to a factoid question answering system for domain specific knowledge bases with significantly simplified language processing techniques.
Files
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access Toh_H_Dissertation_2010s.pdf 6.98 Mb 00:32:19 00:16:37 00:14:32 00:07:16 00:00:37