In Philosophy, Ontology means "the nature and organisation of being"
From modern computer science point of view, it means "shared conceptualization-domain of interest to support communication among humans and computers"
Nowadays, Ontology has been wildly used in many areas in CS, includes (not limit to) AI, Semantic Web etc. However, most of them are created manually.
The previous research about Ontology construction is either based on an existing Upper ontology (to decide whether a property belongs to an ontology or not), or just pre-define an ontology class then using NLP or ML extract information from data source and fill in.
So, don't you think it is a COOL job to make this process more dynamic, accurate and semantic?
We believe the "knowledge" and "structure" represented by a domain ontology could be "transferred" between two closely-related domains.
So, the basic philosophy here is that, in order to create an Ontology about domain X, we start from an existing Ontology about domain Y, which close to X. Then try to find out the internal connectivity between each properties in Y, and WHY they can be a property in Y by analyzing the document about domain Y.
Finally, we construct the Ontology X based on the rules we learned from Y.
Of course, it is never as easy as we describe here, so please keep your on my homepage in the future for more information.