CiviCRM Community Forums (archive)

I have a bit of confusion regarding CiviCRM 4.3 dedupe rules. My understanding is that unsupervised rules need to be more narrow as they run behind the scenes with no oversight and you don’t want them combining contacts that are not actually duplicates. Supervised, as they have manual oversight, can be broader as someone is deciding on a case by case basis if two contacts are duplicates.

An example of each may be supervised consisting of just an email address while unsupervised would be email address, first name, last name. For the supervised rule if the email address found a match then the admin was alerted that a possible match was found. For the unsupervised if ALL fields matched then the contacts would be automatically merged.

Now when I go CiviCRM demo site I see that the unsupervised rule is only matching on email and the supervised rule is matching on multiple fields. Is this not the opposite of what we would want?

I also cannot tell what the field weights are on the reserved rules. I only see the fields but have no idea what weight CiviCRM is assigning to each. Are they weighted?

My concern is that last year for our conference we had to make our default rule ‘fuzzy’ as we wanted one person to be able to register multiple individuals and we ended up with a lot of duplicate contacts. Of course that could very well have been because I used a sloppy fuzzy rule. I’m not seeing a way to set a rule as default in 4.3. I only see options for ‘Unsupervised’. ‘Supervised’, and ‘General’. Is there a ‘Default’ setting and I’m just missing it?

the terminology changed, but the functionality is basically the same.

unsupervised = strict default
supervised = fuzzy default
general = all others

each contact type may only have one supervised and one unsupervised rule -- just as it previously could only have one default strict and one default fuzzy rule. the contact type may have any number of general rules, which are triggered manually during imports or when running a rule across the database to find dupes.

the unsupervised/strict default rule created on installation was always constructed to be just the email address (that dates back for many versions). that was done because that rule is used in the context of contrib forms and event registrations, and the email address is the only natively required field -- any other fields, required or not, are manually added via profile. so basically, that rule was constructed to work with the bare minimum in that context.

I agree that it is too "fuzzy" and that you almost always want to create a more strict rule for use in that unsupervised context. I also understand why the core team decided to make it email-only, as adding any other fields makes assumptions about what data is collected in the forms. there's definitely room for debate.

but you can simply create a new rule and mark it unsupervised in order to replace the system default.