The matchmaking is a crucial operation in Web service discovery and selection. The objective of the matchmaking is to discover and select the most appropriate Web service among the different available candidates. Different matchmaking frameworks are now available in the literature but most of them present at least one of the following shortcomings: (i) use of strict syntactic matching; (ii) use of capability-based matching; (iii) lack of customization support; and (iv) lack of accurate ranking of matching Web services. The objective of this paper is thus to present the design, implementation and evaluation of the Parameterized Matching-Ranking Framework (PMRF). The PMRF uses semantic matchmaking, accepts capability and property attributes, supports different levels of customization and generates a ranked list of Web services. Accordingly, it fully overcomes the first, third and fourth shortcomings enumerated earlier and partially addresses the second one. The PMRF is composed of two layers. The role of the first layer is to parse the input data and parameters and then transfer it to the second layer, which represents the matching and ranking engine. The comparison of PMRF to iSeM-logic-based and SPARQLent, using the OWLS-TC4 datasets, shows that the algorithms supported by PMRF outperform those proposed in iSeM-logic-based and SPARQLent.
展开▼
