Development and Application of Case-based Reasoning Systems based on Semantic Retrieval Agent

Abstract

This study aims to propose an intelligent recommender system to build a more user-friendly environment for various scenarios such as Business-to-Consumer e-commerce or E-learning environments Users can express their requirements in the form of a sentence or short-text rather than exact keywords if recommender system accepts natural language Also the recommender system might reduce the fatigue user experience and help them obtain what they want if it can provide an efficient ranking based on semantic similarity and collaborative filtering This study presents a novel case-based reasoning framework that includes a collaborative filtering mechanism and a semantic-based case retrieval agent In this study we integrate the short-text semantic similarity (STSS) and recognizing textual entailment (RTE) into the case retrieval agent This study compared with existing STSS and RTE methods to evaluate the performance of the proposed approach According to the results the proposed approach outperforms most previously described methods Moreover a case study of an online bookstore was conducted for investigating the effectiveness of the proposed approach The results indicate that the proposed approach outperforms the system using string similarity and a famous e-commerce system On the other hand a potential application using semantic-based approach for e-learning environment was presented in another case study

Date of Award	2017 Feb 13
Original language	English
Supervisor	Tzone-I Wang (Supervisor)