Towards an ontology and DHT-based publish/subscribe scalable system

In the context of Publish/Subscribe systems (Pub/Sub), Peer-to-Peer (P2P) solutions based on Distributed Hash Tables (DHT) offer efficient functionalities such as event routing flexibility, scalability, load balancing and fault tolerance. For event routing, it uses event identifiers computed by event encryption. Likewise, Pub/Sub systems based on ontologies offers well adapted semantic expression capabilities allowing consumers and providers to easily describe services' requests and properties. Thanks to the ontologies, semantic reasoning allows to provide enhanced event discovering. Indeed, DHT-based systems are based on traditional matching algorithms while in ontology-based systems, users can use various literals semantically equivalent but syntactically different expressions in order to describe their requirements. However, both approaches present limitations in terms of: semantic expression restrictions for the first one and scalability issues for the second one. Our solution consists in combining the advantage of both of them, by taking advantage of the semantic capabilities of ontologies at local domain level and the scalable P2P-DHT solutions at inter-domains level. This paper proposes an approach based on a community-wide ontology in order to allow publishers and subscribers to use a common semantic space to characterize production and consumption of resources and services. Based on this ontology, a P2P network based on distributed reasoners (i.e. inference engines) will be able to evaluate if a specific content belongs to an ontology-based community topic. Thus, we use these topics to compute keys for routing events in a DHT.