GaeaPN: A Petri Net Model for the Management of Data and Metadata Derivations in Scienti c Experiments

One important aspect of scienti c data management is metadata management. Metadata is broadly de ned as information about data (e.g., content, source, processing applied, precision). One kind of metadata which needs special attention is the data derivation information, i.e., how data are generated. In our application domain of geographical information systems (GIS) and global change research, we view scienti c objects according to three di erent extents: spatial, temporal, and derivation. Extents are dimensions along which domain users \naturally" access and select information from the database. While the spatial and temporal extents have been studied and formal semantics to those extents proposed, derivation semantics have been ignored. In this paper, we present our own view of what metadata is and propose extensions to current database models to include implicit metadata management. We formulate a de nition of extents as dimensions of metadata, and speci cally discuss derivation metadata and the semantics of such specialized relationships. The proposed model, GaeaPN, is based on a variation of Petri Nets and extends current models such as the object-oriented data model. Petri nets are interpreted as a model to capture and manage data derivation relationships between scientifc data as well as the procedures and algorithms to derive data. We formulate the basic model, study its closure with respect to a \metadata space," as well as its closure with repect to database states. The important contributions of this work are... ... We believe that this framework, useful for GIS and global change studies, generalizes well to other scienti c elds. This work is supported by the National Science Foundation under Contract IRI-9116988.