Semantic Indexing for Complex Patient Grouping

In this paper we describe indexing techniques based on domain knowledge made available in the form of ontologies. In high level interfaces like those used in many data warehousing applications, it is advantageous to use terminology familiar to the enduser of the system. This terminology is very often different from the one incorporated in the underlying database. Much of the terminology (i.e. concepts) in an end-user ontology will map to complex collections of data warehouse attribute value pairs. We create a mapping between the end-user terminology and attribute-value pairs in the data warehouse. We optimize performance by indexing the data warehouse using an ontology. We optimize performance by indexing the whole database using an ontology that represents the end-user’s terminology.