Mining ontologies to extract implicit knowledge
Abstract
With the exponentially growing of data available on the Web, several projects were created
to automatically represent this information as knowledge bases(KBs). Knowledge
bases used in most projects are represented in an ontology-based fashion, so the data can
be better organized and easily accessible. It is common to map these KBs into a graph
to apply graph mining algorithms to extract implicit knowledge from the KB, knowledge
that sometimes is easy for human beings to infer but not so trivial to a machine.
One common graph-based task is link prediction, which can be used not only to predict
edges (new facts for the KB) that will appear in a near future, but also to nd misplaced
edges (wrong facts present in the KB). In this project, we create algorithms that uses
graph-mining (mostly link-prediction based) approaches to nd implicit knowledge from
ontological knowledge bases. Despite of common graph-mining algorithms, we mine not
just the facts on the KB, but also the ontology information (such as categories of instances
and relations among them). The implicit knowledge that our algorithms will
nd, is not just new facts for the KB, but also new relations and categories, extending
the ontology as well.