The physical and logical differences among information sources on the
Internet complicate information retrieval. For instance, data is no lo
nger just simple text or tuples, but now includes objects and multimed
ia. Data can also have varied and often arcane semantics. Sources have
different policies, procedures, and conventions and are hosted by div
erse platforms. Ontologies-models of concepts and their relationships-
are a powerful way to organize query formulation and semantic reconcil
iation in large distributed information environments. They can capture
both the structure and semantics of information environments, so an o
ntology-based search engine can handle both simple keyword-based queri
es as well as complex queries on structured data. Ontology-based inter
operation is especially good at dealing with inconsistent semantics. H
owever, ontologies are difficult to construct. The Java Ontology Edito
r (JOE) helps users build and browse ontologies. It also enables query
formulation at several levels of abstraction. The authors discuss the
use of JOE to develop a health care information system.