eDiscovery questions

• green3dlg

  Newbie

  Points: 5

  More actions

  March 18, 2008 at 7:11 am

  #69626

  I am trying to scan data in our 2000 database for certain words or phrases. I am using a lookup table with those phrases and then using LIKE search clauses to try and find it in multiple fields in our tables. I am trying to take into consideration ignoring if it is just a part of a word (looking for "NT" but ignoring "WANT"). My search string looks like the following: LIKE 'NT%' OR LIKE '%[^A-Z]NT%' Is there something else I might need to get all of the words that are just NT? The fields are all varchar.

  Also, what would you suggest for some fuzzy logic for misspellings? I have tried to use SOUNDEX, but I am getting way too many bad matches.

  Any suggestions would be appreciated. My only tool I'm using is Query Analyzer.

• Gonzalo-603688

  Old Hand

  Points: 384

  More actions

  March 18, 2008 at 7:22 am

  #790794

  Hi,

  I would recommend you to use a %NT% over the fields.

  However, you should check the architecture of the app. Maybe you can make regular searches on the data en build like a catalog. Also, be sure to use Full Text Search functionality

  Regarding the misspelling, you should try to use something like RegEx (regular expressions) and make a B-tree with the different grades of matches.

  It's really hard to implement, that's why only few apps have it.

  Cheers,

  Gonzalo