Cherwell Scientific Publishing 1998
Since we are operating with Merlin - it is hard to think
of something better or more generic then a Spread Sheet.
Although in our case, the flat spreadsheet is just a slice of the multidimensional internal data structure and only holds
a very small proportion of the total data, so we prefer to call it Spread Sheet Viewer.
We decided that drag-and-drop is the most intuitive interface, this does not exclude other methods, but it is used consistently. Also we have avoided both showing all the data (which is large and messy) or hiding any unused data where it will be difficult to obtain. Rather the program shows all available methods/data as tags and leaves it to the user to choose which to view.
In order to make this system flexible each column is a
Java Bean which implements some well defined interfaces.
This allows new methods or visualizations to be readily added by us or users or even third party developers.
A specific example of this is the "custom" expression which allows users to assemble mathematical equation to operate on the data in the data model. These equations are actually stored in the data model itself. This also uses a drag-and-drop interface for consistency.
ss - nice example of predictive caching