Use this URL to cite or link to this record in EThOS:
Title: Managing Hybrid Reasoning on the Semantic Web using a Blackboard System
Author: McKenzie, Craig
ISNI:       0000 0001 3625 1918
Awarding Body: University of Aberdeen
Current Institution: University of Aberdeen
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
In this thesis, we propose a Blackboard Architecture as a means for managing hybrid reas~ning (Le. the combination of ontological, rule- and constraint based-methods) on the Semantic Web. We describe how our S'emantic Web Blackboard coi'Wines the components of . a traditional blackboard system with the principles of the Semantic Web - languages for data representation (RDF and OWL); rules (SWRL) and our own formalism for expressing constraints (CIFISWRL). The traditional blackboard systems used a set of fixed abstraction levels designed specifically for the problem, Iirp.iting the blackboard to a single domain and requiring its contents be-pop~lated according to this pre-determined structure. We discuss our approach of dynamically constiucting an ontology (which plays a similar role to the abstraction levels) on the blackboard at run-time which structures the collated information and allows domain independence. Because the blackboard exists as a shared, semantic resource that is compatible with the surrounding knowledge sources and since inference is one of the fundamental building blocks of the Semantic Web, we introduce the idea of giving the blackboard inference capabilities of its own rather than keeping it as a passive data store. We argue why this is more effective by exploring the . positive and negative aspects of this approach. We identify the key ty'pes ofknowledge source and show how they can be combined to enable reasoning, using differing methods, to be performed over Semantic Web data. We explore the interplay of various knowledge source combinations and detail how their interactions affect the architecture. We put forward the notion of relevancy as a policy for controlling the content of the blackboard. The controller decides what information is placed upon the blackboard and how that data is to be structured. When faced with a potentially very large and diverse dataset it is desirable to only perform inference on as small a sub-set of that information as possible, and not waste effort deriving irrelevant facts. We demonstrate ho'w our Semantic Web blackboard architecture can be employed to achieve this. We show how reasoning over the open world Semantic Web can successfully be applied to closed world problem solving using a combination of negation and the local closed world assumption. The fle~ible configuration of the architecture allows for a more dynamic flow of control during execution than a non-blackboard based architecture.. We provide an example walkthrough of our test-bed system, the AKTive Workgroup Builder and B'lackboard (AWB+B), illustrating the interaction and cooperation of the knowledge sources and providing some context as to how the solution is achieved. We conclude with an analysis of our findings and a discussion of the strengths and weaknesses of the architecture.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available