Generating and Querying Semantic Metadata and Ontologies
Google Tech Talks
June 21, 2007
Semantic technologies aim at providing machine interpretable descriptions of information and services. Ontologies and related metadata provide the technological backbone for these semantic technologies that gain more and more relevance also in commercial settings, for example for information integration, search and retrieval.
The talk will discuss various recent developments in the area of ontology and metadata management. As a starting point we present Semantic Mediawiki - an extension of the well-established Mediawiki platform that runs Wikipedia - for the large-scale collaborative generation of semantic metadata. We also show how simple semantic relations can be automatically generated using methods of pattern-based relation extraction. Subsequently we discuss how more expressive domain models can be generated using techniques of ontology learning from text. Finally we present techniques for querying knowledge bases using natural language interfaces that allow an easy adaptation to new domains. The talk will conclude with an outline of future developments.