From: Julio S. <jul...@gm...> - 2010-11-16 14:35:00
|
2010/11/16 Benny Malengier <ben...@gm...> > > > So, one could instead store basic assertion, like > > person/Name/Surname - implied by - sourcekey > > with the requirement that all assertions follow from a source one way or > the other. > A theory then holds the rules about how one can work with assertions. > Eg, the question if the Name follows from real sources, comes down to > proving > > Is 'Person/Name' 'deductible' from primary sources? > Well, primary sources are often wrong, either by mistake or by deliberate misstatements. I've seen many primary sources I don't believe in because they contradict other primary sources that say something different. They can't all be right, so you have to decide which version you believe or hold you decision till later. > Which one can proof from its subitems. The problem lies in how fine-grained > things must be. Do you need to have assertions for surname, given name, > nick, or do you only want if for the entire name? Answering these questions > allows you to know how a datamodel might have to change. > I think the problem runs deeper. Many genealogy problems amount to: Is person P1 mentioned in document D1 the same person as P2 mentioned in D2? The data model needs to hold duplicate persons whose identity/sameness depends on whether you believe one theory or another. Likewise, the same person but with different parents, children or spouses. Some things can be modelled in GEDCOM, but in general it can only represent conflicting data about elementary facts (I use this extensively), conflicting descents are much harder to model (and document as such). Regards, Julio |