WikiTree Network Corner Cases

+12 votes
287 views

Eva and Bernard asked some good clarification questions about my last post "WikITree Network Defined". Specifically, what do my networks look like for complicated cases, like re-marriages and half-siblings and how do distance measurements compare between these networks. I've compiled a new post to cover this Q&A: https://www.sligocki.com/2021/06/24/wikitree-network-corner-cases.html

WikiTree profile: Space:100_Circles
in The Tree House by Shawn Ligocki G2G6 Mach 1 (16.7k points)

3 Answers

+4 votes

I'll just quickly answer to  the first bit about the Person Network: yes it mirrors the way the connection finder works.

I have a nice case of serial monogamy, which results in a few half-siblings - one step apart in the connection finder - and a few step siblings from either end of the chain - two steps apart in the connection finder. Like this: AS is a brother of AG who is a brother of NG.

* The Connection Finder counts individuals as siblings when they have at least one parent in common. (One parent or two does not matter for the CF).
* The Connection Finder counts marriages when a spousal relationship between two profiles has been entered in the WikiTree database. It's a yes or no, no shades of gray or "it's complicated"

The WikiTree database does not have a separate (redundant) field for a sibling relation. (Imaginably there could be one, but that's not economic, is it?) At least one parent is needed to show people as siblings. If there is only the one shared parent this is not marked specially. If one of the sibling profiles has two parents it shows up as (half), as it does when both have two parents but only one of them is shared.

The Connection Finder ignores the "half" aspect but bases it's one-step sibling relationship on having a common parent.

by Eva Ekeblad G2G6 Pilot (432k points)

So the Bipartite Network models the genetic network, expanded to include the biblical marriage "For this cause shall a man leave father and mother and shall cleave to his wife, and the two shall be one flesh" and extending it to non-married relationships resulting in children :-)

As for the Family Network, yes, I think I understand how it works. It's getting to be an abstraction out of the Wiki-Tree database, fairly different from the abstraction the Connection Finder makes. It is going to be interesting to see what comes out of it.
Haha, yes. I should have called these the Flesh nodes :)
+4 votes
Why do we care about eliminating cliques, if the cliques are real?
by Jonathan Crawford G2G6 Pilot (103k points)
Yeah, that's a good question. Certainly most families are cliques (both I in the Person Network and even in the non-Jargon sense of the word)! However that does not mean that every network model has to have mathematical cliques. That is what I want for my Family Network. Note that it is possible to have cliques even in my Family Network, but it would only happen in strange situations like a person marrying their own aunt/uncle. Specifically, I am trying to avoid these sorts of "automatic" cliques. Where every time a person remarries it automatically forms a clique. Why do I care so much? I think it will make it easier for me to untangle this web and interpret the WikiTree Network more clearly ... but perhaps I am throwing away useful information ... only time will tell :)
Shawn, did you create an illustration of the Brigham Young clique?  I'd be interested to see what it looks like.

Clique 56

Here you go :/

Thank you!

I love it!  Seeing it--even, of course, just a theoretical representation--really adds to my understanding.

Shawn, it's late here but I drop this idea, maybe it does not fly at all, but let's try.

Define P-Node (Parent Node) as the set of one person X and all her/his children (0+)

The P-Nodes of  X and Y are merged iff X and Y are spouses AND the set of their children are equal. In this general case they have the same structure as your famly Nodes.

All the corner cases seem to boil down to have P-Nodes with a single parent, sharing only a part of their children, hence not merged

Put an edge between two single-parent P-Nodes if they have at least one common child and/or if they are spouses.

Et voilà.

Seems to me from my scribblings that does not create more nodes than your Family Node model.

Siblings are grouped in the node(s) of their respective parents in all cases, hence neither artificial distance nor useless clique. Brigham Young sits in a big single parent P-Node with his many children, each of his wives in their own, and only 56 links.

I'll let you make a beautiful drawing of this.

+3 votes

Not sure about the usual meaning of "corner case", but I just checked in the two first circles of Jean Joseph. Among people who ever married, over 20% had multiple marriages. Is that considered "corner"? Think about a corner territory in a Go game covering 20% of the board. That's a BIG corner.wink

by Bernard Vatant G2G6 Mach 5 (56.0k points)
Heh, yeah. Calling them corner cases may be misleading, I agree. These are the complicated cases where slight implementation details will cause different interpretations ... but that doesn't have quite the same ring to it :/

Related questions

+14 votes
4 answers
+12 votes
5 answers
+25 votes
8 answers
+17 votes
5 answers
434 views asked Jun 29 in The Tree House by Aleš Trtnik G2G6 Pilot (567k points)
+27 votes
2 answers
361 views asked Jun 2, 2018 in WikiTree Tech by Paul Gierszewski G2G6 Mach 5 (53.7k points)
+30 votes
8 answers
+11 votes
5 answers
+8 votes
0 answers
75 views asked Jul 2 in The Tree House by Bernard Vatant G2G6 Mach 5 (56.0k points)
+15 votes
3 answers
270 views asked May 18 in The Tree House by Bernard Vatant G2G6 Mach 5 (56.0k points)

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright

...