I get some complaints in this regard, because some people object to Berkeley DB. The fact is that it's the easiest to implement, it scales remarkably well in this and other, larger projects, and most of us already have it. Also, if you don't like the Lingua::Wordnet data delimiters (| and ||), change them with the globals $DELIM and $SUBDELIM in convertdb.pl and Wordnet.pm.