Free Text Searching

I am shortly to embark on a project which will involve the storage of considerable quantities of text data in an oracle database.

I wish to perform free text searches on this data, but am loath to use the LIKE operator because:

1. I suspect it would be terribly inefficient
2. The text data will be stored in a variety of tables/columns
3. I desire more versatile searches than the syntax of LIKE permits.

I have more in mind a scheme in which a preprocessor analyses the text data and generates text indexes, which are themselves held in oracle tables. This would operate similarly to the glimpse/harvest web indexor, but would draw its input from database tables rather than html files.

Has anyone done anything similar? Bright ideas, hints, thoughts, pointers to further reading, hoots of derision, etc, welcome.

Copy to email please.

-- 
    Mike Gahan 
    Information Systems Division
    University College London

    http://www.ucl.ac.uk/~ccaamrg/