Hello. I have used the GEXF export to create a graph in Gephi. It mostly works right, but I have a lot of unconnected people or separated trees - people which I can verify are connected to my main tree. I have the impression that it is linked to families without children in which the “spouse” link is not present.
I don’t know if the problem is in the export or in Gephi: @Nick-Hall (I understand you wrote this module) have you seen this before?
I just checked the .gexp export, and it seems that it is what I thought: some (if not all - my database is too large to check completely) families without children are not correctly exported. The “spouse” link is not generated, creating isolated individuals or sub-trees in the graph.
1 Like
It’s a problem with the export.
I didn’t realise that anyone was using this export. I wrote it as a proof of concept for someone who was interested in using graph database to analyse Gramps data.
The problem is that the export only looks at parent families of every person when writing the edges. It needs to consider all families, but then it also needs to be careful to avoid writing duplicate edges.
Thanks for the answer.
In fact,I have always wanted to manipulate my genealogy as a graph, and other exports from Gramps (SQLite, CSV) are quite complicated to use because of the structure with multiple lookups.
I have used yED with GEDCOM exports, but the manipulations as not as easy as in Gephi once you have understood how it works. So when I found it, I spent a whole afternoon playing with it - and found the described issue.
If ever you have time to correct the export, I’d be very grateful (but it’s really not a matter of life and death 
)
[edit] This morning I had the idea to use yED as a translator from GEDCOM to GraphXML and it seems to do what I want - I’ll explore more later.
You might try this GEDCOM to GEXF converter. I have not tried it myself, but the code looks very simple and appears to create only people nodes, with edges for father and mother (no family nodes). It looks like it would include nodes for people who have no parents.
1 Like
Thanks, I’ll take a look
That is an interesting repository by Guillaume Jacquenot.
The Pythagoras tree python sub-repository looks like it could be massaged into a simple compressed fractal visualization of the completeness of a pedigree with the shape.
The Pythagoras Tree image, with smooth coloring, making it more resemble a real tree. by Atze van der Ploeg Atzecsse
Perhaps a false color spectrum could be used to visualize attributes… like source or children count. Or lifespan.