Multicriteria analysis of ontologically represented information

Our current work concerns the development of a decision support system for the software selection problem. The main idea is to utilize expert knowledge to help the user in selecting the best software / method / computational resource to solve a computational problem. Obviously, this involves multicr...

Full description

Saved in:
Bibliographic Details
Main Authors: Wasielewska, K, Ganzha, M, Paprzycki, M, Bădică, C, Ivanovic, M, Lirkov, I
Format: Conference Proceeding
Language:English
Subjects:
Analytic hierarchy process
Computation
Data structures
Decision making
Hierarchies
Multiple criterion
Ontology
Regression analysis
Software
Structural hierarchy
Support systems
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Our current work concerns the development of a decision support system for the software selection problem. The main idea is to utilize expert knowledge to help the user in selecting the best software / method / computational resource to solve a computational problem. Obviously, this involves multicriterial decision making and the key open question is: which method to choose. The context of the work is provided by the Agents in Grid (AiG) project, where the software selection (and thus multicriterial analysis) is to be realized when all information concerning the problem, the hardware and the software is ontologically represented. Initially, we have considered the Analytical Hierarchy Process (AHP), which is well suited for the hierarchical data structures (e.g., such that have been formulated in terms of ontologies). However, due to its well-known shortcomings, we have decided to extend our search for the multicriterial analysis method best suited for the problem in question. In this paper we report results of our search, which involved: (i) TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), (ii) PROMETHEE, and (iii) GRIP (Generalized Regression with Intensities of Preference). We also briefly argue why other methods have not been considered as valuable candidates.
ISSN:0094-243X
1551-7616
DOI:10.1063/1.4902284