Using semantic mappings for semantic based publish/subscribe systems

Abstract

Routing of information within heterogeneous, distributed network domains (e.g. communication networks, ubiquitous computing environments) is a key challenge that must be tackled for such environments to be successful. Increasingly, it is accepted that such routing systems need to cope with mobile and volatile sources and destinations of heterogeneous information, and Semantic-based Publish/Subscribe (SBPS) systems have been proposed as a means to support this. However such systems typically assume a common semantic model underpinning the routing. Devices, network resources and applications within network domain environments are likely to be heterogeneous, as they will be produced by different vendors who will each have different priorities and design philosophies. It is unlikely that a single standardized semantic model for communication will gain widespread adoption and the need for exchange of heterogeneous information will persist. The SBPS systems must allow applications/SBPS routers to use different semantic models to communicate with each other without the necessity of custom building gateways. Finally the system must support flexible, adaptive semantic interoperability to cater for the dynamics of such environments. The KBNMap system proposed in this thesis uses an adaptive, multi-strategy mapping service approach (comprising a set of mapping strategies and adaptive strategy selection mechanism) for the purpose of supporting multiple semantic models within a SBPS, in support of routing heterogeneous information over a network in an appropriate and adaptive manner. The mapping strategies are used for efficiently loading semantic mappings for use with distributed and heterogeneous knowledge-based applications. The probabilistic-based strategy selection mechanism enables the system to dynamically select the appropriate mapping strategy, in order to adapt to the dynamics of the environment where aspects of the semantic models, applications and network environment are constantly changing. This research advances the state of the art by introducing and evaluating a multi-strategy mapping approach and adaptive mapping selection mechanism within a SBPS system, including identification of the key factors that will influence strategy selection.