My general research interests are in the emergence of the Semantic Web and more specifically in the in the fields of Artificial Intelligence, Machine Learning and Information Retrieval.
My PhD work has been focused on the problem of Automatic Ontology Mapping, i.e. finding semantic correspondences between similar elements of different ontologies.
Ontology is a formal explicit specification of a shared conceptualization of a domain. Sample ontologies vary from the simplest glossaries and data dictionaries, to thesauri and taxonomies, XML schemas, and formally expressed ontologies. Ontologies have been suggested as a way to solve the problem of information heterogeneity in the Web by providing formal and explicit definitions of data, and reasoning ability over related concepts. Given that no universal ontology exists in the real world, ontology mapping arises as an intriguing research problem in the area of both AI and Semantic Web areas for years.
Ontology mapping can be tackled either by hand or using automated tools. Manual mapping becomes impractical as the size and complexity of ontologies increases. Full or semi-automated mapping approaches have been examined by different research studies. Representative approaches include analyzing linguistic information of elements in ontologies, treating ontologies as structural graphs, applying heuristic rules and machine learning techniques, and using probabilistic and reasoning methods etc.
My approach is based on the state
of the art, but excels them by integrating propagation theory, information retrieval techniques,
machine learning algorithms and artificial intelligence model. The
breakthroughs of the approach include:
The evaluation of the approach has been executed in
Ontology Matching campaign 2006 and 2007, by the Ontology Alignment Evaluation
Initiative. The preliminary results are promising and appear to be scalable.
I am also interested in exploiting Semantic Web technologies to support information retrieval tasks and improve Web search results. Simple keyword queries are valuable when people know exactly what they are seeking for and/or the information is well-defined. However, simple keyword searches does have the limitation of returning false relevant documents containing homonyms and missing truly relevant documents that have only synonyms or abbreviations. Meanwhile, ontology based search can overcome the problem by utilizing its built-in inference engine to perform deduction or other forms of automated reasoning to locate documents that are semantically similar to the query terms regardless of the surface difference, thus achieve more accurate or comprehensive results. Furthermore, since ontology-based search provides a set of access points over the whole ontology, associated recommendation can be implemented via relationship within ontology nodes. Such ideas have been implemented in the DiLight system, an ontology-based information access system for e-learning environments.
In the future I am interested in the following research directions: