Research Of Matching Technology In Content-based Publish/Subscribe

The application range and scale of distribute system has been changed considerably with the fast development of Internet. Recently, the distribute application may expand thousands of millions of nodes, and also include more and more mobile entities. These entities exist in the world widely; their location and behavior may have a lot of variety during the periods, they need to correspond and cooperate to achieve a collective purpose. These restriction shows that we need a flexible communication model, reflecting dynamic and loose–coupling hypostasis of the application. The content-based publish/subscribe considers as the best system structure, and bring widely attention and investigate. The publish/subscribe model based on event-alternate has loop-coupling abilities in space, time and synchronization, and also provides one-to-many and many-to-many asynchronous communication model. We do a lot of study in matching arithmetic which is one of the key technologies in publish/subscribe system, and the main content is following below:First we expatiated for the basic data model of event and subscribe language about publish/subscribe system, and also introduced some typical routing style in publish/subscribe system, including routing based on global knowledge, filter identity, filter covering, filter merging, and using of advertisements. After then it discussed some typical attribute filter, defined conjunctive of filters, covering of conjunctive filters, identity and overlapping of conjunctive filters and merging of conjunctive filters formally.It put forward a mutil-index matching arithmetic, discussed the structure principle of matching and achieves it. It canceled typical counter and used the mutil-index matching algorithm to find suitable events quickly, reduced repeat matching, and then increased the capability of system effectively.Last we put forward a fuzzy matching arithmetic supporting subscription covering, constructed the fuzzy vectors, and built the matching rules. It also defined the membership function of attributes and the threshold to judge whether the matching is success or not, implemented the fuzzy matching arithmetic and did the experiment. We need not to do precision matching for non-precision chooses. The algorithm filtrates unsuitability attributes to find the publish event quickly which matches the subscription basically.