For a real blast from the past, the ‘rdbms’ application included word seach indexing support, based on the Porter Stemming Algorithm, courtesy of Hans Nilsson.
My interest at the time was to try to figure out how to expand Mnesia’s indexing support so that something like that could be automated.
A few of the ideas later made it into Mnesia in the form of index plugins (undocumented, but super-cool). Weighted indexes, supported by rdbms, aren’t in Mnesia, though.
I also recall playing around with hierarchical indexes (indexes of indexes), but I think that was actually in ErlHive, which was even more out there. Not even going to link to it.