ECML PKDD 2008

Semantic Unification in a Real-World Data Flow for Delivering Business Intelligence

MDC Partners is based in Belgium and delivers business and market intelligence to pharma and medical device companies. As heterogenous public databases and the internet are our main data sources, semantic unification of concepts is the necessary driving force behind our data gathering en data storage platforms. In this talk we discuss a number of design philosophies and technical challenges in building a semantic data gathering, mining and reporting engine. Specifically on the semantic unification side, we will discuss alternative schemes for semantisizing text fragments, based on bootstrapped probabilistic grammars, reference ontologies and factual semantic data. The impact on the total semantic engine, from incoming data to reported queries, of alternative approaches will be discussed.