Partial Matchmaking for complex Product and Service Descriptions.
Heiner Stuckenschmidt, Martin Kolb · MADOC (University of Mannheim) · 2008
Abstract: Matchmaking between offers and requests is an essential mechanism inelectronic market places. Description Logics have been proposed as a appropriate fra-mework for representing offers and requests and standard methods from descriptionlogics reasoning have been proposed for computing matches. We argue that classicalreasoning is not well suited for realistic scenarios as it does not allow to compute par-tial matches and qualify the mismatch. We propose a more flexible approach for mat-ching complex product and service descriptions based on the notion of approximatesubsumption in expressive description logics and present a prototypical implementati-on of our matching system. 1 Motivation Asmoreandmorebusinessisperformedonline,theimportanceofelectronicmarketplaceswhere products and services are traded in a semi-automatic way is increasing steadily.One of the basic problems connected to an efficient use of electronic marketplaces is theproblem of matchmaking between offers for products or services and requests by potentialcustomers [Vei03]. Matchmaking tries to find offers that as closely meet the requirementsofarequesteraspossible.Thistaskisnon-trivialasinmanycases,itisnotpossibletofindperfect matches. In this situation, the matchmaking algorithm has to find ’good enough’offers. Determining which offers are ’good enough’ often does not not only depend onoffer and request it self, but also on the preferences of the customer and the intended useof the product or service. This means that matching has to be customizable towards thespecific customer.Another challenge for successful matchmaking in electronic marketplaces is the need tointegrate heterogeneous product and service descriptions. As the representations used bydifferentparticipantsfortheirinternalpurposesareoftengearedtowardsthespecificneedsof the company and show little standardization across different enterprises. In order to beabletousethisinformationinmatchmaking,openstandardsforclassifyinganddescribingproducts have been developed that provide a common framework for describing productsand services by assigning products to product classes and by defining a number of pro-perties for describing the concrete offer in more details. Examples of such standards areeClass, UNSPSC and RosettaNet [HLS08].It has been argued that matchmaking in electronic markets can benefit from the use of