ijcai.pdf

Ateret Anaby-Tavor and Avigdor Gal Technion – Israel Institute of Technology good automatic semantic reconciliation algorithm would rank the e xact mapping relatively close to the best mapping, thus enabling an ef?cient search of the exact mapping [1]. 2 Preliminaries We start by introducing the notions of attribute similarity measure and mapping similarity measure. The interested reader is referred to [13], where we have grounded these no- tions in a theoreticalmodel of uncertainty based on fuzzy sets. To illustrate our model we shall use simpli?ed schemata of two car rental reservation systems, as follows: AvisRental(RentalNo, PickUpLocationCode, PickUpDate, PickUpHour, PickUpMinutes, ReturnDate, ReturnHour, ReturnMinutes, Price.) AlamoRental(RentalNo, PickUpLocation, PickUp-Date, PickUpHour, PickUpMinutes, DropoffDate, DropoffHour, DropoffMinutes, Price.) It is worth noting that this example was largely simpli?ed, for the purpose of clarity. In Section 4 we provide our empir- ical analysis, which is based on data that was collected from multiple Web sites. 2.1 Attribute similarity Given two attribute sets  and   , we associate a similarity measure, normalized as a similarity degree between(total dissimilarity) and(equivalence), with any mapping among attributes of  and  . Therefore, given two attributes