Information Technology Reference
In-Depth Information
Fig. 8.6 Overview of constructive induction process whereby mapping between database ele-
ments as described via their metadata and corresponding ontology concepts are used to induce new
“facts” concerning the contents of the database. In this general case, concepts 6-8, which is
included in the ontology but does not map to the database construct, is used as an intermediate
concept to defi ne a concept triplet or higher order construct involving multiple intermediate enti-
ties that begins and terminates with data elements that map to concepts in the ontology construct
￿
Phase 1 - Metadata to Conceptual Knowledge Entity Mapping : In the fi rst
phase of implementing a CI-based agent, the metadata that serves to defi ne a
knowledge source of interest (e.g., a data dictionary or equivalent description of
the contents of a data set or sets) must be mapped using either manual or auto-
mated processes to the entities that comprise one or more conceptual knowledge
collections (e.g., syntactic or semantic matching of metadata defi nitions to enti-
ties in a terminology, ontology, or equivalent construct). This process usually
results in one-to-many mappings, in which each metadata items corresponds to
more than one conceptual knowledge entity. For example, if mapping a clinical
data set with the specifi c variable corresponding to a “White Blood Cell Count”,
depending on the mapping approach being used and the intent of the KE initia-
tive, that variable could be linked to multiple ontology-anchored concepts, such
as the molecular entity “White Blood Cell”, the laboratory procedure “White
Blood Cell Count”, as well as the clinical fi ndings of “White Blood Cell Count
Normal”, “White Blood Cell Count High”, and “White Blood Cell Low.” This
process generates a “knowledge map” that resolves individual variables of
interest in the metadata being utilized to a corresponding set of atomic concep-
tual knowledge entities.
Search WWH ::




Custom Search