Research And Implementation Of Publish/Subscribe System Based On "Scientific Knowledge Ontology" On P2P

With the rapid development of Internet, mobile devices and broadband networks, network computing becomes more and more complex. The main characteristics of the new generation of network computing are large-scale, decentralized control, dynamic, and autonomous. Since publish/subseribe systems have the advantages of loose coupting, anonymity, many-to-many communication and scalability, they have become important infrastrueture middleware platforms support for the new generation of network computing.However, the existing pub/sub systems can not match event and subscribe according to semantics of event, and most is C/S mode system. For this reason, we introduced Ontology into pub/sub systems, designed and implemented a pub/sub systems based on "Scientific Knowledge Ontology " on P2P. The system uses "Scientific Knowledge Ontology " to express the concept model, combined P2P with Web Service to implement the network system, made sharing the field of scientific research into practical.This paper presents a new type of pub/sub systems data model, we express the event in SK-PS system with semantic vector form. It is called Event Model, which includes Class Vector and Property Vector. The subscription model of SK-PS system is built on the basis of event model, used to assign the expression and filter form of subscription. Therefore, user's subscription conditions actually is a composition of certain elements, which established on the event semantic vector, equally, we use the idea of semantic vector to build model for subscription conditions of system.Based on this new type of data model, this paper given an efficient matching algorithms, called matching algorithm based on semantic vector index. The algorithm is to complete to find all subscribers conditions from subscription model, which is suited to the item of event's semantic vector index model, if an event match to the conditions of a subscription, it should meet three conditions: The property set of subscription is subset of event property set; There has one class at least in all classes of event, which is Sub-category of the type in subscription conditions, equivalent or same to it; the property value in event vector is corresponding to the value of same property in subscription vector.Finally, we realized SK-PS system in .Net environment.It verified data model and matching algorithm, fully showed the characteristics of the P2P networks and autologous practicality at the same time.