A Methodology for Creating Ontology-Based Multi-agent Systems with an Experiment in Financial Application Development

Abstract

The creation of an ontology for a particular business domain has advantages for application development, interoperability, reusability, and integration. Agent systems with an underlying ontology can exploit these advantages by providing semantically-aware applications. However, there is a lack of support for ontology development in existing methodologies for designing and building multi-agent systems for business applications. We describe a methodology and an experiment for building such systems. The approach focuses on the development of ontology as the driving force of the development processes and strives to put application development in the hands of domain experts, requiring as little as possible from software developers. We describe two complementary aspects of the approach: (i) a phased, iterative methodology and (ii) a pipeline of select tools with which to carry out the methodology. We report results of an experiment in developing an application for finance, including an evaluation, lessons learned, and recommendations.

The Ontology-Based Methodology Phases To Develop Multi-Agent System (OmMAS)

“…Nowadays there are many methodologies that have been proposed based on a variety of concepts, notations, and techniques [4,5,8]. Among them are the ontology development methodologies, such as MOMA [4], Methontology [10], Kactus [11], On-To-Knowledge [12], Enterprise [13,14], Cyc [15], Sensus [16], Tove [17], NeON [18], Diligent [19], (KA)2 [20], and CO4 [21,22].…”

“…Moreover, developing ontology knowledge is not an easy process and need more time, resources and deeper analysis [22][23][24]. This paper focuses on the methodology for developing Ontology-Based Multi-Agent System (MOMA) [4] as the current ontology-based methodology in a multi-agent system. This paper investigates in details the ontology development phase of the MOMA.…”

“…The objective of multi-agent systems is to build distributed, intelligent, and robust applications [2]. However, there are many issues faced by the researchers to develop the multiagent system, such as the issues of structured and unstructured data; heterogeneity of the system and information; interoperability between systems and many more [3][4][5][6][7]. Currently, one of the prevalent issue in multi-agent system is the semantic aspect [4].…”

“…Even though there are many updates regarding the new methodology techniques to handle semantic aspect on multiagent system, discussions on the ontology-based in the methodology phase is still limited [4,8]. One critical problem that makes the development of multi-agent system becomes unsuccessful is because of not using a good development methodology [2,5,9].…”

“…The aim of this paper is to critically analyze the existing development methodology on multi agent system, named Methodology for Developing Ontology-Based Multi-Agent Systems (MOMA) [4]. After identifying the limitations of MOMA, this research proposes an improved development methodology named Ontology-Based Methodology Phases to Develop Multi-Agent System (OmMAS).…”