Research On Web Services Discovery And Selection Based On Semantic Matching

Semantic web service, which means applying the technology of semantic web to the fieldof web service, is able to make web service models in the ontology level and to cooperatebetween services in the semantic level. The semantic-based web services discovery meansfinding out services which meet the needs of users, such services describes by semanticdescription framework. Web service has become the mainstream technology ofService-Oriented Computing because of its flexibility, character of cross platform andtransportability. A titanic service assembly comes to being because of the situation wherenowadays, services from web have been growing explosively. So web services may come outwith similar or the same aspects, such as functionalities, not-functionalities，preconditions andeffects. It is a pivotal problem to locate satisfying services from such a titanic servicesassembly immediately.Currently，the precision and recall of service selection method are relatively low, becausethe traditional web services remain based on the level of syntax and key words in order todiscover web services. Semantics-based web services discovery is able to overcomedisadvantages of traditional web services lacking of semantic information. It searches bymeans of using a domain ontology or a large-scale vocabulary database, improves theprecision and recall rate of services discovery to a certain degree by means of adopting alogical reasoning mechanism, however, as causes low efficiency of a discovery at the sametime. It is generally defined a service matcher, so far, as is one of methods to studysemantics-based services discovery. Some of service matchers simply match semantics afterdefining semantic matching level, while others take actions to make up for the lack ofmatching after semantic matching failure.In this paper, we propose a services discovery mechanism based on similarity of semantic matching and semantic context of information in order to improve the precision andrecall rate of web services discovery. Furthermore, we propose a relatively completeframework of web services discovery based on the mechanism.Main jobs of the framework are as followed:(1) the research on Web ServicesDescription Language. Web services must use the same description language once they arediscovered. Researched and analyzed service description languages such asSAWSDL,WSMO,OWL-S,we adopt OWL-S as services description language. We modelservices in the semantic level using the ontology constructed by OWL and define four levelsfor semantic matching. Then, based on the jobs above, can we define the mechanism ofservices discovery to match by services semantics.(2) non-reasoning-based match forsemantic web services. So far, the inefficiency of semantics-based web services discoverymainly caused by using the method of logical reasoning to discover web services. We canimprove the efficiency by means of adopting binary coding to determine the relationshipbetween the two concepts (ancestor-child, child-ancestor).(3) A services discovery based oncalculating the similarity of semantics of information. After the failure of semantic matching,in order to return more services related to the users’ requests, we propose an algorithm, whichis based on calculating the similarity of semantics of information, to make up for the negativeeffects caused by simply semantic matching.(4) We have implemented a prototype usingXAMPP as web server, select owl-tc2.0as our test suite, adopt Java as our developmentlanguage. Results of the experiment indicate that framework we proposed above not onlyimproved the precision and recall rate, but improved the efficiency of services discovery aswell.