摘要：UDDI is a universal description, discovery and integration protocol. As a public registry of Web service, it is designed to store information about each company and its service. Traditional centralized service discovery structure of UDDI service registration center does not apply to large-scale service discovery. When all the services register to a center, the service bottleneck, failure of single point and the poor scalability defects will occur. In addition, traditional service matching mechanisms are mainly based on keywords method which lacks of semantic description and makes the service publisher and demanders cannot reach a common semantic understanding. This will lead to the problems of semantic conflicts and low accuracy that seriously affects the precision and recall of service matching. To address these shortcomings of the centralized service discovery structure of UDDI, we propose a distributed semantic service registration center which is in the construction of loosely coupled P2P network enabled the progressive massive search. In the P2P distributed network, there can be a large number of nodes to store the registration information which is suitable for large-scale service because of the adaptivity, scalability and good fault tolerance characteristics. In order to reduce the number of concurrent transmitted packets, the advanced ant colony algorithm is introduced to forward packets by probabilistic choice. The results comparison with the traditional algorithm is given through the simulation experiments and it has shown that the proposed method has good performance for the distributed service discovery
