BioNavigation was the first
prototype developed in collaboration with Louiqa Raschid and Maria Esther Vidal,
University
of
Maryland
,
College Park
.
The motivations of the projects were to develop a system that would
support resource discovery as
well as cost prediction. The
characteristics of BioNavigation are the following.
·
The resources stored in the semantic map are interlinked data sources;
·
The ontology is a simple graph where each node represents a scientific object (e.g.,
gene) and an edge between two nodes denotes the relationship between two
scientific objects;
·
Each resource is mapped to the scientific object(s) it provides information
about (e.g., GenBank is mapped to the ontology
class Sequence);
·
Each link (index, hyperlink, etc.) linking two resources is an instance of the edge
linking the two corresponding scientific objects;
·
The cardinality of each data source as well as the
domain, range, and cardinality of links are stored as metadata in the semantic map;
·
Resource discovery queries are path expressions on the ontology;
·
Users may select a semantics to predict the cost of physical path (e.g., return
the resources more likely to return as many entries as possible);
·
Answers of a resource discovery query are a list of
physical paths ranked with respect to the selected semantics.
