Forum Discussion

TTree2007's avatar
TTree2007
Level 4
7 years ago

Classification - UK PII

Hi Rod

As briefly mentioned, having a lot of msg's being returned running classification for 'UK PII'.  The customer is treating them as "false positives" as it does appear to be practically capturing all email msgs!  

Firstly, can you clear up how the UK PII Policy should trigger the tag please?  I'm reading it as it needs to have...

1) 'Any of' - 2 or more - DOB, email address, UK postal address.

2) 'Any of' - CC, UK Driving License, Electoral Reg, NHS no, Nation Insurance no, passport no, Tax ref, etc...

My understanding is that BOTH these conditions need to be met for it to tag it as "PII" but it appears by reviewing a few tagged files and testing an msg file it does seem to trigger PII with just 1) - e.g. my test msg have 3 email addresses and 2 UK postal addresses.

There was NO elements from 2).

If so, this seems a bit simplistic as practically every email that's been forwarded or replied to has these elements!  I'm hoping you have a simple fix or I'm missing something. 

I haven't put the latest CHF on these environments (it's using base 6.1).

Thanks

  • Rod_p1's avatar
    Rod_p1
    7 years ago

    We have released the 6.1RP1 / 6.1.1 Rolling Patch which includes the previous hotfixes for the PII information templates and other Veritas Information Classifer (VIC)  enhancements. I suggest you do your modifications and changes on that release for the newest available version.

     

    Rod

4 Replies

Sort By
  • andrew_mcmullan's avatar
    andrew_mcmullan
    Level 3
    7 years ago

    It's 1 OR 2 (not 1 AND 2).  Note the root 'Any of' that joins 1 and 2.

    But also note that the email address condition requires 10 or more email addresses.

    So a document with 3 email addresses and 2 postal addresses should not get a match.

     

  • andrew_mcmullan's avatar
    andrew_mcmullan
    Level 3
    7 years ago

    I suggest changing the confidence on the email condition to (very high to very high) instead of the default (high to very high) as this will then require extra keywords to be present intended to identify email addresses used in documents (non email) vs just email addresses in email (too common).

    and/or increase the number of email addresses required to 1000. 

    • TTree2007's avatar
      TTree2007
      Level 4
      7 years ago

      Thanks.  I have been playing with confidence level.  We decided to add some further conditions and a reg ex and have it working quite nicely now!

      Regards

      • Rod_p1's avatar
        Rod_p1
        Level 6
        7 years ago

        We have released the 6.1RP1 / 6.1.1 Rolling Patch which includes the previous hotfixes for the PII information templates and other Veritas Information Classifer (VIC)  enhancements. I suggest you do your modifications and changes on that release for the newest available version.

         

        Rod