Basic NLIDB

A basic NLIDB can be achieved by TFIDF or simple search on concatenated fields or inverted index and w determiner. Add a little bit of logic for functions, implicit entity and refining as designed by me and this will beat quite a few NLIDBs. I wonder why no one has proposed this solution. Can also make it ngram based. Joins can also be handled. Add graphs to make it awesome. If multiple columns contain similar values, use frame semantics or frame triggers to separate them. This will work especially with poor grammar and lack of column names. Am I a genius or what?? Accept me as your king. Accept defeat. Bend your knee. Pledge your allegiance.