In browsing the current Data.gov holdings, one sees data which has in some instances been chunked (for example large datasets which would contain too many records as CSV) - perhaps these may be broken out by state, et cetera. Similarly, there are cases where there are datasets which are part of a time series (e.g. 2005->2006->2007 annual data releases) - it would be useful toward usability to have ways to relate these and treat them as groups, e.g. treating an entire collection of individual state files as a unit, or being able to navigate (for example, if I am looking at the 2005 Tennessee dataset, I might want to be able to quickly jump to the 2006 Tennessee dataset).
Idea No. 37