Edinburgh Napier University

  Taxonomists classify organisms in order to refer to, identify and understand them. Over time, the same organism may be classified according to different taxonomic opinions. As alternative classifications multiply, taxonomists are faced with the need to compare and contrast taxonomies to identify how they differ. A taxonomy is hierarchical, and multiple taxonomies with overlap result in a network or graph structure. Representing and manipulating graph structures has limited support in traditional database technology. This paper reviews support for graph manipulation in existing databases and discusses their limitations with regard manipulating taxonomic information. The authors describe the design and implementation of an extended graph database model that supports a semantic differentiation of relationships, type genericity, and instance synonymy. A query language has been developed which is a pattern matching language that avoids explicit specification of graph traversing queries by the user. A prototype has been developed in Java and an OODBMS (POET).