|
OntoMiner: Mining
& Populating Ontologies from the Web
|
Team
Hasan Davulcu
Srinivas Vadrevu
Saravanakumar Nagarajan
Josh Ryan
Fatih Gelgi
Vish Ramachandran
Hung Nguyen
|
|
Abstract
Ontologies
provide a mathematically sound conceptual formulation of domain knowledge
and capture complex relationships between objects in a domain. Ontologies
are becoming increasingly popular in understanding and encoding the
knowledge in various areas, including bioinformatics, natural processing
and formal languages. They
found applications from building a meta search engine to automatic semantic
tagging of documents. In order
to widespread usability for the Semantic Web, there is a need to bootstrap
large, rich and up-to-date domain ontologies that organize most relevant
concepts, their relationships and instances. Currently many of the existing
ontologies have either been manually encoded or semi-automatically generated
with human feedback. The goal
of the OntoMiner project is to mine and populate specialized domain
Ontologies by transforming and mining a set of overlapping Web sites that
provide views of information from the same underlying domain. The experimental results on News and
University Web sites indicate that this is a promising approach to
automatically extract ontologies from domain-specific Web sites.
The
OntoMiner project aims to develop novel web mining techniques, that detects
and exploits regularities within Web sites, to turn them into the so called
"Semantic Web" representations. Such representations enable
creation of conceptual and perceptual maps of specific areas of interest, as well
as operators for analyzing content to allow users to explore the Web of
facts and relationships to discover trends and unusual instances.
CIPS Internal Link
|
