[TSC-public] Are personae top-level entities?
ttw4 at verizon.net
Thu Sep 4 18:32:00 CDT 2014
On Sep 4, 2014, at 6:28 PM, Calum MacLean wrote:
> In this case, I'm only talking about evidence-level persons and events.
> I would tend to favour the multi-level persona model (though I haven't thought it through fully). At both evidence and conclusion level, persons would have the same structure (more or less); correspondingly for events.
> At evidence level, the persons would be associated with a source.
> At conclusion level, a person would refer lower level persons (normally evidence level, but not always in multi level case). So a conclusion level person would be made up from a number of evidence level persons.
This is the model that I and a few others have been pushing for some time. There is the two-tier approach where person records appear at two levels, as conclusion persons that refer to evidence persons, and evidence persons that refer to sources. There is also an n-tier approach where person records appear in a tree -- the root of the tree is the final conclusion person; the leaves of the tree are the evidence persons, and the intermediate node persons are partial conclusions. A benefit of this n-tier approach is that the reasons for the final combining together of evidence, which by its nature is a statement made up of tree-structured clauses, is implicit in the structure of the tree.
In order to enable this, all one needs to do is allow person records to have a new type of link to other person records. For example, these evidence and conclusion structures could be added to GEDCOM by adding a "1 INDI @xx@" line to the "0 INDI" record. That's a lot of power added for what looks like a trivial addition.
> On 4 Sep 2014 23:56, "Richard Smith" <rsmith at fhiso.org> wrote:
> On 04/09/14 21:40, Calum and Norma MacLean wrote:
> Another option could be that the source *contains* persons and events:
> person 1
> name: Jimmy Smith
> person 2
> event 1
> type: birth
> role: person 1
> This seems natural to me, as you find the names of persons and the
> events within the source.
This is the Behold approach.
> That would seems natural if persons and events can *only* appear within the context of a single source. As Enno says elsewhere in this thread, that's not an unreasonable requirement at the evidential layer.
> But do we want the same person objects and event objects to represent conclusion-layer objects? The conclusional objects don't belong under a single source. There wasn't complete agreement in the "How do you manage your genealogical evidence" thread, though if you adopt the multi-level persona model favoured by Tom and me, there's no difference between an evidence-layer persona and a conclusion-layer person. If you adopt such an approach, it would perhaps seem odd to place some person and event objects in sources, and others outside them.
More information about the TSC-public