Recent Research Interests
I am currently investigating techniques to provide Description Logic (DL) reasoning sevices over dynamic
data. This includes providing more efficient incremental classification, realization, query
answering, etc., under updates to DL knowledge bases. I am particularly interested in applying these
reasoning techniques to content syndication on the Web.
Syndication on the Web has attracted a great amount of attention in
recent years. As technologies have emerged there has been a transition
to more expressive syndication approaches; that is subscribers (and
publishers) are provided with more expressive means for describing
their interests (resp. published content), enabling more accurate
dissemination. In particular, there has been a transition from keyword
based approaches to XML and more recently to using RDF, which is the
underlying representation format of RSS 1.0.
Recently, there has been very recent interest in using OWL ontologies for syndication, providing an even more expressive representation of subscribers interests and published content. One of the main benefits of OWL is the support for formal reasoning in one of its sub-languages OWL-DL, whose semantics are firmly founded in Description Logic (a decidable fragment of First Order Logic). In such an approach Description Logic reasoning is used for content dissemination providing accurate matching of more expressive subscriptions. While DL-based syndication provides increased expressivity over XML and RDF based approaches, it is evident from initial efforts that the approach suffers from scalability issues. One of the main limitations is reasoning under changing data. This is primarily due to the static nature of existing DL reasoning techniques. Additionally while there has been a variety of work in proposing distributed architectures in non-DL syndication settings, such approaches have not been addressed in the DL context. Rather, all previous work has focused on having all reasoning done by a central broker (reasoner).
The general goal of my current research is to address the previously mentioned
limitations in DL-based syndication approaches and to create a
framework that allows practical application of DL reasoning to
syndication on the Web. I am currently working on extending the
incremental reasoning techniques I have previously developed. I am
also developing practical belief revision algorithms for OWL-DL knowledge
bases, as this service will be necessary for syndication purposes
because information is often conflicting. I plan to apply the
reasoning techniques to a unified syndication
framework; in order to address scalability issues, I am
investigating the distributed aspects of such a syndication
architecture.
Some overview slides related to my research are available here.
Please also feel free to look at my publications for some related papers.
Past Research Projects
Masters Thesis (Under the direction of Dr. I Budak Arpinar and
Dr. Amit P. Sheth)
Abstract:
The focus of contemporary Web information retrieval systems has been to provide
efficient support for the querying and retrieval of relevant documents. More
recently, information retrieval over semantic metadata extracted from the
Web has received an increasing amount of interest in both industry and academia.
In particular, discovering complex and meaningful relationships among this
metadata is an interesting and challenging research topic. Just as the ranking
of documents is a critical component of today's search engines, the ranking
of complex relationships will be an important component in tomorrow's Semantic
Web analytics engines. Building upon our recent work on specifying and discovering
complex relationships in RDF (Resource Description Framework) data, called
Semantic Associations, we present a flexible ranking approach which can be
used to identify more interesting and relevant relationships on the Semantic
Web. Additionally, we demonstrate our ranking scheme's effectiveness through
an empirical evaluation over a real-world dataset.
Full Text
Presentation
Note: Evaluation details are available upon request.
Past Projects Involved In:
SemDis - Semantic Discovery:
Discovering Complex Relationships in Semantic Web
Ranking Semantic Associations
Semantic Association Identification
and Knowledge Discovery for National Security Applications
TRAKS - Terrorist Related Assessment
using Knowledge Similarity
|