> It will be huge, but I don't see a problem with it, you will need to
> look at indices statistics after the table is populated to see if you
> have a lot of duplicates, if so, you could add the primary key to the
> end of the table to help garbage collection process to run.
>

If he only adds unique words as has been described, there will be no
duplicates

;)

> I'd like to know your progress and how well it works and the gotchas

you

> find.

Ditto,

I imagine dealing with (ignoring) common words that introduce false
positives, limiting search results for vague queries, and coming up
with a ranking system that works in the real world will be the
challenge. Even a feedback loop from the UI about particular results
to create a self tuning ranking system.

Adam