An attempt for a classification / From Doug on Obstap-list Dataset (the generic dataset == ObsDM or Theory/Solar/Planetary data) Table Catalog Simulation (etc. - SimDAP and the like) Project Project_NRAO_VLA Project_NRAO_VLA_Proposal (for tables this can get pretty open ended) Image Image_Cube Image_Cube_Spectral Image_Cube_Time Image_Cube_Polarization (etc.) Image_Longslit Image_Grism Image_Mosaic Image_Simulation (etc.) Spectrum Spectrum_1D TimeSeries (light curves are only one type of time series) SED (not sure about subclassification here) Event Event_Rosat (etc.) Visibility Visibility_NRAO_VLA Visibility_NRAO_ALMA (etc. Note that at the root we have the generic Dataset which is where we get global data discovery of any type of data. If we look for table, image, spectrum, etc. we get all subclasses. Not shown are instrumental data collections which might be extensions such as like Spectrum_1D_KPNO_Coude or Visibility_NRAO_VLA. I would argue that most astronomical data in our archives can fit into such a classification scheme - but we need more use cases. There are some cases, e.g., single dish radio data, for which it is not clear where the instrumental data would fit. The suggestion here is that if we look at the data in our science archives, most of it fits within such a classification somehwere. This is similar in some respects to what Pat suggested where we classifiy by number of axes, except that that is not sufficient to distinguish between data such as images and tables. This is pretty rough and needs more thought for the finer classifications, but the question for now is how close this might come to classification of actual data. - Doug