GraPS: A Graph Publish/Subscribe Middleware

Pub/sub is an elegant paradigm for disseminating information efficiently and anonymously among producers (publishers) and consumers (subscribers). However, with current topic and content-based pub/sub approaches it is difficult to formulate subscriptions that adequately and accurately express the interest of the consumers in a semantic information domain. In this paper we introduce GraPS, a pub/sub middleware that provides a publication model based on graphs. Points of interest in the information domain are mapped to nodes, and relationships between points of interest are mapped to edges. Consumers can effectively express their interest in publications by means of graph sub script ions that exploit the properties of nodes and the semantics of the edge relationships. Graph subscriptions do not require complete knowledge of the graph and can be updated whenever the consumer's interest changes. Furthermore, graph subscriptions are automatically updated whenever the information domain changes. We illustrate GraPS by means of three application scenarios and present a set of experiments with an implementation of GraPS based on standard pub/sub middleware.