Tom Wetmore wrote: > On Monday, May 23, 2011 7:47:11 PM UTC-4, Steven Gibbs wrote: >> Basically, I really only have two types of object, persons and documents. I >> input a document and create a new persona for each name in the document. >> Then I check each persona to see if I am comfortable about merging it into >> an existing person %>< >> I didn't cater easily for unmerging persons. If I need to unmerge a person, >> I have to create a new person for every original persona in the documents >> for a person, unlink any parents and children and then remerge each of the >> personas again. If I was starting again, I'd keep a record of each merge so >> I could recreate the last two persons that were merged, as experience shows >> that it's usually the most recent merge that was in error. > Very nice. I'm working on a similar solution. I handle the unmerge problem > by never really merging. Instead of merging I build up a tree of person > records. If I decide two personae refer to the same person I create a new > person record that simply refers to the two personae. I can add a > justification to that new record to explain the rationale for joining. If I > decide later that the personas refer to different people, I just delete the > higher level person. > Take it a step further. Have a separate entity for your "higher level person" and then link entities between that and the personae records. At a minimum the link could simply contain pointers in each direction but you could expand it to contain a note and maybe some value to indicate your confidence - including negative values to say you think the persona does not refer to that person. Gramps, BTW, does have an undo for merges but some operations, such as importing a fresh GEDCOM, wipe the history. -- Ian The Hotmail address is my spam-bin. Real mail address is iang at austonley org uk

On Tuesday, May 24, 2011 4:38:24 AM UTC-4, Ian Goddard wrote: > Tom Wetmore wrote: > > On Monday, May 23, 2011 7:47:11 PM UTC-4, Steven Gibbs wrote: > >> Basically, I really only have two types of object, persons and documents. I > >> input a document and create a new persona for each name in the document. > >> Then I check each persona to see if I am comfortable about merging it into > >> an existing person > %>< > >> I didn't cater easily for unmerging persons. If I need to unmerge a person, > >> I have to create a new person for every original persona in the documents > >> for a person, unlink any parents and children and then remerge each of the > >> personas again. If I was starting again, I'd keep a record of each merge so > >> I could recreate the last two persons that were merged, as experience shows > >> that it's usually the most recent merge that was in error. > > > Very nice. I'm working on a similar solution. I handle the unmerge problem > > by never really merging. Instead of merging I build up a tree of person > > records. If I decide two personae refer to the same person I create a new > > person record that simply refers to the two personae. I can add a > > justification to that new record to explain the rationale for joining. If I > > decide later that the personas refer to different people, I just delete the > > higher level person. > > Take it a step further. Have a separate entity for your "higher level > person" and then link entities between that and the personae records. > At a minimum the link could simply contain pointers in each direction > but you could expand it to contain a note and maybe some value to > indicate your confidence - including negative values to say you think > the persona does not refer to that person. > Ian Ian, Thanks. I have described my current thoughts for "doing genealogical research" as building up "person-trees" with personae (person records codified directly from evidence records) at the leaves and higher level person records as the roots and "interior nodes" of the trees. This is a bit of a simplification since it ignores issues of events codified from evidence, levels of confidence, adding notes, adding conclusions, and so forth. My entire model is a bit more complex that the simple tree of persons idea, but these concepts seem subtle enough that I hesitate to obfuscate the core ideas with that complexity. I guess I'm trying to say that my overall model does include the points you have just mentioned. Thanks. Tom