Aljuboori, AS, Coenen, F, Nsaif, M and Parsons, DJ 2018, 'Performance of case-based reasoning retrieval using classification based on associations versus Jcolibri and FreeCBR : a further validation study' , Journal of Physics: Conference Series, 1003 (1) .
|
PDF
- Published Version
Available under License Creative Commons Attribution. Download (609kB) | Preview |
Abstract
Case-Based Reasoning (CBR) plays a major role in expert system research. However, a critical problem can be met when a CBR system retrieves incorrect cases. Class Association Rules (CARs) have been utilized to offer a potential solution in a previous work. The aim of this paper was to perform further validation of Case-Based Reasoning using a Classification based on Association Rules (CBRAR) to enhance the performance of Similarity Based Retrieval (SBR). The CBRAR strategy uses a classed frequent pattern tree algorithm (FP-CAR) in order to disambiguate wrongly retrieved cases in CBR. The research reported in this paper makes contributions to both fields of CBR and Association Rules Mining (ARM) in that full target cases can be extracted from the FP-CAR algorithm without invoking P-trees and union operations. The dataset used in this paper provided more efficient results when the SBR retrieves unrelated answers. The accuracy of the proposed CBRAR system outperforms the results obtained by existing CBR tools such as Jcolibri and FreeCBR.
Item Type: | Article |
---|---|
Additional Information: | Ibn Al-Haitham First International Scientific Conference, 13-14 December 2017, Baghdad, Iraq |
Schools: | Schools > School of Computing, Science and Engineering |
Journal or Publication Title: | Journal of Physics: Conference Series |
Publisher: | IOP Publishing |
ISSN: | 1742-6588 |
Related URLs: | |
Depositing User: | USIR Admin |
Date Deposited: | 26 Jul 2018 10:06 |
Last Modified: | 15 Feb 2022 23:34 |
URI: | http://usir.salford.ac.uk/id/eprint/47921 |
Actions (login required)
![]() |
Edit record (repository staff only) |