Use case request: Simple GPS activity

[nilsbrummond]

I saw a request for use cases somewhere in the www.gedcomx.org website. So here goes:

Front page of the wiki (https://github.com/FamilySearch/gedcomx/wiki) first line:

GEDCOM X defines an open data model and an open serialization format for exchanging the components of the genealogical proof standard.

I'm having a hard time figuring out how the results of a simple GPS activity will be model in GEDCOMX based on the current conceptual model spec or #182.

First what is a good simple sample GPS activity for a basic use case? How about "John Smith [HYPOTHETICAL CASE]" page 36 of "GPS - Building a solid Case" by Rose.

Question: was John Smith the son of William Smith?

Five steps of GPS:

1. All records known to be relevant where searched.
2. Evidence and sources of these..
   • source1 (county birth register), evidence: William as the father of a John who was born 3 March 1872.
   • source2 (contemporary family bible), evidence: gives same parentage and birthdate for John.
   • source3 (1880 census), evidence: lists William Smith household contains John, aged eight years.
   • source4 (letter from John smith), evidence: dated 1894 , which begins "dear father" is addressed to William Smith.
   • source5 (1906 death certificate), evidence: for John Smith gives the same birth date as above and list William Smith as the father.
   • source6 (Gravestone of John Smith), evidence: calls him "son of William".
3. analysis of the data show William Smith as John's father...
4. No conflicts to resolve
5. Conclusion of William as John's father

What would the object model look like to describe the outcome of this basic GPS activity?

[jralls]

After #182 is finished and merged, I'd do it like this:

• 6 SourceDescriptions, one for each source -- perhaps more to account for hierarchies, transcriptions, and what have you, but let's keep it to the 6 enumerated (SD1 -- SD6).
• An EvidenceAnalysis (EA1) document with SourceReferences (SR1 -- SR6) linking it to each of the 6 sources, in which you would explain the chains of evidence supporting the birthdate and the father-son relationship.
• A SourceDescription (SD7) pointing to the EA1, and a SourceReference (SR7) pointing to that.
• Two Persons, one each for John (P1) and William (P2), with SR7 in the sources list of each.
• A birth Event, with EventRoles pointing to the P1 & P2 and SR7 as the source.
• A Relationship, again with the P1 and P2 with SR7 for the source.

[EssyGreen]

1. Is not really covered by GEDCOMX (see Mapping GEDCOMX to the process model #141)

2-3 I can't comment in detail since I've lost track of what's what or what/where the latest model is. John's example sounds about right but personally I would want some form of evidence link between each of the Persons and each of the relevant SourceDescriptions before I attempted to analyse the whole caboodle in a single "EA1"

4-5. Are assumed to be within the verbatim text of the evidence link(s)

[jralls]

personally I would want some form of evidence link between each of the Persons and each of the relevant SourceDescriptions before I attempted to analyse the whole caboodle in a single "EA1"

The evidence analysis is the evidence link: It's the proof argument that demonstrates that the evidence applies to the Persons. Having it in between sources and the conclusions is the major feature that puts GedcomX in compliance with the GPS. It can also describe what has been searched and what was or was not found, thus covering the requirement of a "reasonably exhaustive search".

[EssyGreen]

Yes I know that John but your example implied that there would be a single which linked all the sources to the person ... all I was saying is that I (as a researcher) would personally also want to have pairings of a single source/person.

It can also describe what has been searched and what was or was not found, thus covering the requirement of a "reasonably exhaustive search".

That's a bit like saying you can shove anything in the text so therefore it supports anything.

I don't believe GEDCOM X really provides any features to help the researcher define, organise or analyse their goals, searches or search results.

[nilsbrummond]

@EssyGreen

That's a bit like saying you can shove anything in the text so therefore it supports anything.

I don't believe GEDCOM X really provides any features to help the researcher define, organise or analyse their goals, searches or search results.

I agree.

And give the first line of the wiki page:

GEDCOM X defines an open data model and an open serialization format for exchanging the components of the genealogical proof standard.

I think GEDCOM X needs to support all of the components of the GPS before it can be called complete and meet the definition on the wiki.

[EssyGreen]

I think GEDCOM X needs to support all of the components of the GPS before it can be called complete and meet the definition on the wiki

+1

[jralls]

That's a bit like saying you can shove anything in the text so therefore it supports anything.

Says the woman who's argued extensively for shoving everything into text. ;-)

[jralls]

I don't believe GEDCOM X really provides any features to help the researcher define, organise or analyse their goals, searches or search results.

Defining, organizing, and analyzing goals goes a bit beyond the GPS and while it would make an attractive feature in a genealogy program, I'm not sure that it belongs in a standard aimed at exchanging results.

Search is another matter, at least to the extent of documenting what's been searched, and I can see an automated use case: Interacting with programs like Gensmarts. Why don't you open an issue and propose a structure?

[EssyGreen]

Says the woman who's argued extensively for shoving everything into text

Context is everything ;) I actually argued for allowing free form text instead of having to code up every different type of fact etc. That's a bit different to just lumping chunks of the whole process into a text field :)

Defining, organizing, and analyzing goals goes a bit beyond the GPS and while it would make an attractive feature in a genealogy program, I'm not sure that it belongs in a standard aimed at exchanging results.

My bad - I (wrongly) assumed we were talking about the process - I forgot GEDCOM X is only interested in the proof standard

Search is another matter [...] Why don't you open an issue and propose a structure?

I did, many moons ago - see #141

[jralls]

I did, many moons ago - see #141

Meaning ToDo?

I think that got lost in a rather wide-ranging discussion. It does bleed a bit into goal-setting, but maybe you should try again with it as its own issue.

[EssyGreen]

I agree it got lost but I'm not sure beating people over the head with the same stuff by re-creating new issues will make it any more likely to be accepted. I prefer to tackle it from another angle ... like this one here :)

[nilsbrummond]

My first take at an object diagram from how I understand things now.

Note: The extract evidence should be optional.

Questions:

1. Every Document will need to have a SourceDescription object paired with it?
2. Is another Document and SourceDescription object needed between source1 and it's extracted evidence?
3. Should GPS1-Source and GPS1 objects be swapped in position, so that GPS1.sources={source1-source6} and GPS1-Source.sources={GPS1}, Remove GPS1 from the GPS1-Source.exctactedConclusions, and GPS1-Source.about={GPS1}?

[nilsbrummond]

New object diagram show "yes" answered to my last three questions...