Since more and more services are published on the open web, service discovery is becoming a critical technology for service applications. In order to discover services it is important to calculate the similarity between services. Current approaches, such as ontology concept-based techniques, and information retrieval-based techniques only consider service's input/output as a simple datatype for service similarity calculation. However real-world services published on the web always have input/output parameters with complex datatype. Therefore it is significantly important to discover services considering complex datatype. In this paper, we propose an algorithm considering complex datatype of service's input/output by mapping services to tree structures for service discovery. And we calculate service similarity using the tree structure. First, we propose a method to mapping services to tree structure considering complex datatype. Second, we use the combination of depth among the trees to calculate the similarity between two service trees. Finally, an experiment is done and our evaluation shows our approach for service discovery based on tree structure has a better performance comparing with approaches which are only considering simple datatype.
展开▼
