From ImageWeb

dia:tag:=Semantic Matching

• Huimin Zhao, Semantic Matching Across Heterogeneous Data Sources, Communications of the ACM, Vol 50 No 1, Jan 2007.

This article discusses the problems of semantic reconciliation of heterogeneous data sources. It focuses mainly on relational data sources, but the discussion is also relevant to Semantic Web sources. I found two main areas of interest:

1. Characterization of "Schema level" and "Instance level" correspondences. While we have been aware of this distinction, it hasn't been very clear in our presentations of what we hope to achieve with data webs, and, in particular, we have not articulated that different techniques might be applicable for the different levels.
2. A survey of machine learning techniques applicable to semantic matching. I'm not sure that any of this is yet for us, but it is probably important to be aware of machine learning approaches to finding correspondences between diverse sources.