Expressive Syndication on the Web Using a Description Logic Approach

Abstract

Syndication on the Web has attracted a great amount of attention in recent years. However, today's state-of-the-art syndication approaches still provide relatively weak expressive power from a modeling perspective and provide very little automated reasoning support. If a more expressive approach with a formal semantics can be provided, many benefits can be achieved, including a rich semantics-based mechanism for expressing subscriptions and published content and automated reasoning for discovering subscription matches not found using traditional syntactic syndication approaches.

In this dissertation, I develop a syndication framework based on the Web Ontology Language (OWL), which is the standardized language for representing the semantics of information on the Web. One of the main advantages of the framework is its support for formal reasoning, as the semantics of subsets of OWL are founded in description logic (a decidable fragment of first-order logic). Therefore, the previously mentioned benefits can be achieved using description logic (DL) reasoning.

However, the main limitation in using OWL as the underlying representation model is related to the overhead of DL reasoning under changing data, which makes the approach impractical for many real-world domains and publication frequencies. Given this, in this dissertation, I develop incremental DL reasoning algorithms for the required reasoning services in the framework. Specifically, I present incremental consistency checking techniques, as well as algorithms to perform more efficient incremental query answering.

Lastly, to demonstrate the practicality of the syndication approach, I have implemented a prototype of the framework and performed extensive empirical evaluations using synthetic datasets, as well as real world data from the financial domain. These results show the effectiveness of the incremental reasoning services and the practicality of the syndication framework in general.