[TSC-public] Progress on ELF / _UID

Albert Emmerich Emmerich.Albert at t-online.de
Sun Oct 29 13:31:40 CDT 2017

We use the _UID to bring data back to the original record after the data had travelled through other applications, the data being modified or updated. I myself do it very often, as I am working with some teams where every member has its own computer / database to work in. The data exchange is done by GEDCOM.

Handling of _UID works in 99% of all cases, if the programs support _UID. Therefore supporting _UID is an important criterium for us when deciding which program should be used.

More often than separating records you have the situation that you merge records, as you have enough evidence they are describing the same person. Because of tracability and to be able to later separate the records again if needed, it is important that all _UID will be kept when merging records. If you make sure you have access to the original data you can separate records later using the _UID.

I admit using the _UID is not 100% fool prove. We had situations where researchers decided to replace a record carrying some no longer needed data by data of another person - using the old record and by this using the same _UID. If two researchers do this independently using the same "old" record, you will have two totally different persons with the same _UID.
So coincidence is a very strong indication for same person described by the records, however to be sure you need additional checks. As I see it you cannot make sure that an abuse as described will happen, and therefore you will never be 100% sure when only using _UID (or any other identification!). However _UID is the strongest indication I know to identify same persons in data.


-----Ursprüngliche Nachricht-----
Von: TSC-public [mailto:tsc-public-bounces at fhiso.org] Im Auftrag von Richard Smith
Gesendet: Sonntag, 29. Oktober 2017 18:53
An: tsc-public at fhiso.org
Betreff: Re: [TSC-public] Progress on ELF / user defined tags

Searching through my rough list on ideas for ELF, I see I've noted a second and arguably more serious concern with _UID.  Presumably its purpose is to provide a globally unique identifiers for records so that they can be identified and merged if they are exported and reimported, but I've never seen that stated definitively software vendors who uses _UID.

If this supposed motivation is correct, it only works if there are agreed semantics on the ownership of UUIDs.  Suppose I send a copy of my work to another researcher.  Some time later I notice an error.  What I thought was one person was in fact two, so I delete some facts from the individual already in my tree and create a second individual with them. 
This may happen over a period of time if I realise there is an error some time before working out what exactly is wrong.  Perhaps the researcher I sent my work to reaches a similar conclusion but splits the record in the other way, so now I have UUIDs X and Y, while the other researcher has X and Z, however my X corresponds to his Z and his X to my Y.  We need to prevent this, and, so far as I know, there is no current understanding in place that does so.

Richard Smith,                       FHISO   <http://fhiso.org/>
FHISO Technical Co-Coordinator       One Community, One Standard

More information about the TSC-public mailing list