Obtaining an optimal MAS configuration for agent-enhanced mining using constraint optimization

Publication Type:
Conference Proceeding
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012, 7103 LNAI pp. 46 - 57
Issue Date:
Filename Description Size
Thumbnail2010006759OK.pdfPublished Version255.76 kB
Adobe PDF
Full metadata record
We investigate an interaction mechanism between agents and data mining, and focus on agent-enhanced mining. Existing data mining tools use workflow to capture user requirements. The workflow enactment can be improved with a suitable underlying execution layer, which is a Multi-Agent System (MAS). From this perspective, we propose a strategy to obtain an optimal MAS configuration from a given workflow when resource access restrictions and communication cost constraints are concerned, which is essentially a constraint optimization problem. In this paper, we show how workflow is modeled in the way that can be optimized, and how the optimized model is used to obtain an optimal MAS configuration. Finally, we demonstrate that our strategy can improve the load balancing and reduce the communication cost during the workflow enactment.
Please use this identifier to cite or link to this item: