Matchmaking Algorithm Research Of Semantic Web Service Discovery

With the development of SOA and the increasing of the amount of Web Services, the quality of Web Services is irregular. It is the focus that how to find the best matched service among the different Web Services. Nowadays, UDDI is usually used to describe, discovery and integrate the Services. But UDDI is based on keywords to discovery services without supporting semantics. In this case, we will miss many services, so add the semantics to UDDI means a lot.Semantic Web will give the resources clear meaning. The computer will identify the semantics of the information automatically, then judge and understand the relationship between each other. Once the idea of combining the Web Service with the Semantic Web was brought, it was accepted by the industry quickly. It is popular to describe semantics based on ontology theory. How to use the relationship between the ontology and the functional and nonfunctional properties to develop the efficient matchmaker is the point of the research. This paper will combine the logic-based and the unlogic-based matching to develop the matchmaker.This paper is based on the theories of Web Service, Semantic Web, ontology theory. Analyze the current Web service matchmakers in terms of IO, PE and Qos. Use different matching methods according to their characteristics. Improve the algorithm of match degree calculation. Design a multi-choice matchmaker based the algorithm. Design the category matching level to filter the unrelated services. Then design four kinds of matching as IO matching, IOPE matching, IOQos matching and IOPEQos matching. Put fuuzy theory to Qos nonfunctional property matching. At last get the sum of the matching degrees of the hybrid matching.IO matching is the point of the matchmaker. It is the important parameter. This paper combines the semantic distance-based method with the attribute-based method. Consider the part-of, is-of and synonymy relationship and gives them different weights. Convert the semantic distance into matching degree with the decreasing function. This paper puts forward a new common function to calculate the matching degree and proves it right and better than the formers with data. Semantic distance-based method depends on the architecture of the ontology network too much, so this paper uses attributes-based method to make up the defect. This paper puts forward a new function to calculate the attribute matching degree to make sure the values between 0 and 1. Put weights to semantic distance-based degree and attribute-based degree and calculate the IO matching degree.PE matching is a referenced matching in this paper. If the user has chosen the IOPE matching or the IOPEQos matching, PE matching is carried out firstly.Qos matching will be focused more and more with the increasing of the Web services. This paper analyzes the factors of Qos attributes and considers service cost, service runtime, service reliability and service availability as the main parameters.At last this paper proves this hybrid matchmaker is effective by experiments. It can get high precision and recall ratio in most cases.