Ontology-based Context-aware Service Discovery in a Globally Distributed Network

Knarig Arabshian
Columbia University


Abstract

In order to accomplish context-aware service discovery, static or dynamic service information must be made available globally and represented in such a way that registering and querying for them can be more complicated than what is available to us now. Imagine a service discovery system where a user can enter the following searches: "Find me a nearby Italian restaurant that has available seating for 5 people" or "Find me a Chinese restaurant or something similar to it that has a nearby movie theatre playing an action movie". This requires efficiently distributing dynamic service information which require frequent updates as well as enhancing querying and registration beyond text and key word search so that specific services as well as logically similar services can be searched for. I have designed and implemented GloServ, a global service discovery system which aggregates different types of services in a globally distributed network. A partial list of such services include: real-time event-based services, location-based services, communication, e-commerce or web services. GloServ improves on current service discovery systems by scaling across a globally distributed network and allowing intelligent querying and registration of services by classifying services in the Web Ontology Language Description Logic (OWL DL). First, I present a novel algorithm, which achieves distribution of service information, by mapping the service ontologies onto a scalable hierarchical peer-to-peer network. Second, I present an enhanced querying mechanism for service discovery which combines ontology queries with key word search.