Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Implementing an open reference architecture based on web service mining for the integration of distributed applications and multi-agent systems

: Kehagias, D.D.; Tzovaras, D.; Mavridou, E.; Kalogirou, K.; Becker, M.


Cao, L.; Bazzan, A.L.C.; Gorodetsky, V.; Mitkas, P.A.; Weiss, G.; Yu, P.S.:
Agents and data mining interaction. 6th International Workshop on Agents and Data Mining Interaction, ADMI 2010 : Toronto, May 11, 2010. Revised selected papers
2010 (Lecture Notes in Computer Science 5980)
ISBN: 978-3-642-15420-1
ISBN: 978-3-642-15419-5 (print)
ISSN: 0302-9743
International Workshop on Agents and Data Mining Interaction (ADMI) <6, 2010, Toronto>
Conference Paper
Fraunhofer IESE ()

This paper introduces an open reference architecture that enables seamless integration of distributed applications and agent-based systems. The reference architecture has been designed in order to be holistic, open and standards-abiding. It is based on an ontological framework that operates as a middleware between application developers and service providers on the one side and multi-agent systems on the other one. The proposed architecture enables seamless integration of closed (standalone) applications, provided that these applications export their functionality in the form of public web services. Moreover, we propose a data mining framework that operates on web service data and performs classification of web services and their operations into their semantically described counterparts. This enables seamless integration of applications through the corresponding web service interfaces, based on the ontological framework. On the other hand we show that knowledge derived from web service mining can be used by multi-agent systems in order to provide composite functionalities derived from the distributed web service operations. Moreover, agents are capable to provide personalized services that fulfill user requirements, by taking into account the personal profile, as well as the history of the end user. The paper presents the prototype tools that have been implemented for the realization of the proposed reference architecture.