Research On P2P Based Semantic Web Service Discovery

Web service technology is not only an important solution to solve the problem of software reuse, but also a low coupling solution for complex heterogeneous software system. With the continuous development of web service technology, massive web services are provided in the open network. Although the huge number of web services has brought great convenience to the industrial production, a higher requirement on performance for service discovery system need to be satisfied, namely make a fast response to the service request with all satisfied web services. Moreover, this also leads to a quality problem that is how to select the appropriate ones from so many web services which provide same function but each has different quality.These days, many unsolved problems are still existing in the web service discovery system, such as the single point of failure, inefficient resource locating methods, inefficient method of network resource utilization, low accuracy of service matching methods, lack of the support for QoS information screening and maintenance. In order to solve these problems, a P2P network based semantic web service discovery framework with QoS information filtration, named SWSDI, is proposed in this paper. Based on the Chord network model and the JXTA grouping method, all nodes in SWSDI are grouped into two logic layers; through utilizing the two phase matching method, most unsatisfied services are filtered out; by using the Chord routing model, service requests are sending to the target nodes quickly. All of these designs make it possible for SWSDI to solve the problems of the existing web services discovery system. An improved semantic similarity matching algorithm is put forward in this paper, which can provided the support for computational service matching and semantic matching of natural language based web services. To support the QoS information filtering, a complete intelligent process of QoS information collecting and maintaining is designed. This paper also took the user history data into consideration:a history management module is implemented based on user preferences, which is used to optimize the service discovery process, namely shorten the response time of the service discovery. A resource management module, which can help super node to schedule QoS validation and select shadow node, is implemented by collecting the available resources of nodes. Utilizing this module can not only realize the full use of network resources, but also improve the stability of the service discovery system. In this paper, two experiments are designed to verify the accuracy&stability of concept semantic similarity matching algorithm, and the precision&recall of SWSDI system separately. The first experimental result shows that the improved algorithm performs good stability&accuracy, and is superior to10other algorithms; the second experimental result shows that SWSDI system provide a high matching result on both the recall and precision within test collection OWLS-TC4.