Poster by: Ian Wood, The Australian National University. 

We present the design and prototype-implementation of a web-based system to support collaborative scientific reasoning based on experimental evidence. At the lowest level, the system comprises databases of "raw" experimental data. The organisation of this data will reflect the structure of the scientific experiments and associated tags would be based on control parameters used for these experiments but it is not otherwise assumed to be a relational database system. As "meaning" becomes associated with datasets by comparison with theory and simulation, metadata tables are constructed dynamically as indexes into the raw data. These metadata tables are dynamic and are hosted by central metadata servers as well as in "collaborative islands" which support particular groups of researchers. At the highest level, a wiki interface supports logical inference based on a dynamic ontology of concepts (and their instances) derived from, on the one hand, scientific theory, and, on the other, archived data and metadata.

Computational support for our system has been built using Java-Jini network technology and and the RacerPro automated reasoning tool, and has been applied to the management of distributed data from fusion energy research. The system supports dynamic metadata, single-sign-on security, web-based and stand-alone browsers for waveform data and the collaborative islands described above.