sources, transcriptions, spelling mistakes, deductions

[paul womack]

There seems to a underlying, unspoken theme in the recent discussions
on variant/nested data.

Many of the examples raised mention that multiple sources
might well spell a name differently, or that a place might
be known by multiple names.

Further, place of birth might well be stated to different
degrees of precision on multiple sources
(I've had a case where the "true" location was probably
a village called "Greater Blah", there also being a Lesser Blah", but one
census simply listed "Blah", and yet another listed the nearby
larger town).

And yet the person was (obviously...) only born in one place.
The difficulty here (I believe) is that there is no clean
distinction between a "raw" transcription, (probably textual)
a semantic transcription (e.g. where the *name* of a place
is interpreteted to a place *object*), an aggregate
interpretation (e.g. where multiple sources are conglomerated
into a consensus place object), and deductions (where the process
of reaching a conclusion involves logic and interpretation
as well as "averaging").

I think some of the recent suggestions as to features
are all tiptoeing around the edges of this very fundemental
concept - that the process of genealogy involves
a progression from "raw sources" to a "concluded tree".

Of course, at the moment, most of us are using
gramps to hold BOTH, (along with the intermediate stages
of the process), and yet there is nothing explicit
in the gramps model (or any other software I'm aware of)
to codify or mark these stages.

I feel that feature development in terms of data representation
should have this distinction (raw->conclusion) in mind as a key guiding design concept.

BugBear

[bugbear]

paul womack wrote:
> There seems to a underlying, unspoken theme in the recent discussions
> on variant/nested data.

Wrong list (obviosuly).

Nothing to see here :-)

BugBear