Author(s): N.I. Hachem, N. Serrao, M.A. Gennert, and K. Qiu Computer Science Department, Worcester Polytechnic Institute, Worcester, MA 01609

Source: WPI-CS-TR-94-3

Abstract: An important aspect of scientific data management is metadata management. One kind of metadata which needs special attention is data derivation information, i.e., how data are generated. In this paper we propose extensions to current database models to include implicit metadata management.

We introduce and analyze Derivation Nets, an extension of Petri Nets, as a tool to represent and manage data derivation relationships between scientific data, and the procedures and algorithms that derive data. We formulate the Derivation Net model, study its properties with respect to database states, and discuss the semantics of queries and updates within this framework.

Derivation Nets have been used to design the metadata manager portion of the Gaea scientific database management system. Their operational characteristics, implementation, and sample use are described.