APG-L Archives

Archiver > APG > 2004-01 > 1072995213


From: "Jerry Fitzpatrick" <>
Subject: RE: [APG] Re: Name Registry?
Date: Thu, 1 Jan 2004 16:13:33 -0600
In-Reply-To: <1d0.16fdf61b.2d25e9b3@aol.com>


Donn,

Thanks for your insights.

Technically, yes, the identifier would identify what the NGS GENTECH model
calls a 'persona'. However, the reason the persona concept exists is to
support cases where there is insufficient evidence to decide whether the
named personas are actually the same individual or not.

Each genealogist already makes a judgment as to whether data about a named
person/persona should be attributed to one individual or another. It makes
no difference whether the person/persona has a unique identifier. For
example, although a SSN appears on some records, it doesn't appear on every
record pertaining to a particular individual. For those unmarked records, a
decision has to be made whether the data applies to the individual whose SSN
we know or a completely different individual.

It's not necessary to create a different persona identifier for each piece
of evidence collected. Indeed it would be very silly to do so. There is a
one-to-many relationship between assertions and a specific persona. For
example, I do *not* need create a different persona (and associated
identifier) for each piece of information I wish to attribute to my
great-grandfather. The evidence is not that ambiguous.

When the evidence is meager and/or ambiguous, the right thing to do is to
assume multiple individuals (and therefore separate personas). I'd argue
that this is the minority case, along the ambiguity no doubt increases as
you delve further into the past.

Regardless of whether I'm right or not, the sheer number of personas doesn't
concern me. You have either defined the personas or not. I have enough
unique identifiers for all of them.

Numbering aside, statistics from the Social Security Administration or any
large study would be very interesting and enlightening. Such statistics
could easily prove or disprove my hypothesis that name, birth date, and
birth place are very unique. In either case, this is a good example of the
kind of quantitative analysis that could be done using genealogical data.

---
Jerry Fitzpatrick
Software Renovation Corporation

http://www.SoftwareRenovation.com


-----Original Message-----
From: [mailto:]
Sent: Thursday, January 01, 2004 3:23 PM
To:
Subject: [APG] Re: Name Registry?

Jerry:

I think you reached the heart of the matter when you wrote:

> A globally unique identifier is basically a convenient surrogate for a set
> of real-world data (e.g. name, birth date, etc.). If that set of data is
> duplicated but you suspect two separate individuals are involved, you can
> create different identifiers for each.

What you seem to be describing is not a unique Identifier for an individual
person (who somehow would have to be distinguished from other individuals
with
similar names through other identifying characteristics in order to be
assigned a unique identifier), but rather a unique identifier for a persona,
as that
term is used in the Gentech Data Model--a statement or data set that
includes
a name and associated data or factual information, as given in or by a
particular source.

The number, then, would not identify a particular individual or person, but
rather a set of data associated with a name from a particular source. A
unique
identifier for that data set could serve a useful purpose in data exchange,
but when you consider the number of personas associated with any individual
person--the number can reach hundreds--the potential size of the registry is

staggering.

Assignment of a unique persona number would still leave to each genealogist
the judgment call on whether any given persona, whether or not uniquely
identified by number, should be attributed to a particular human person, or
whether
inconsistencies among possibly related personas suggest instead the
existence
of two or more different individuals to whom they should be attributed.

For most individuals living in the US since the early 1900s, their Social
Security Account Number will serve to uniquely identify them as persons.
Before
that, unique identity remains problematic. Social Security has asked for
full
name at birth, date and place of birth, and names of parents (and more
recently, parents' Social Security numbers) to describe and distinguish
individuals
applying for Social Security account numbers. Now that it has the
information
from the SS-5 application forms in a database, it would be interesting to
know
how frequently different individuals have had the same date and place of
birth and names of parents, up to the time parents' Social Security numbers
were
first required. That would give us some indication of the extent to which
those
data items uniquely describe an individual.

Donn Devine, CG, CGI
Wilmington DE

CG, Certified Genealogist, CGI, and Certified Genealogical Instructor are
service marks of the Board for Certification of Genealogists, used under
license
by board certificants after periodic evaluation, and the board name is
registered in the US Patent & Trademark Office.


==== APG Mailing List ====
The Association of Professional Genealogists
http://www.apgen.org/publications/apg-l/index.html

---
[This E-mail scanned for viruses by Declude Virus Scanner]



This thread: