[4 / 2 / ?]
Quoted By: >>51014108
Hey /g/
TL;DR: how would you visualize an extend family tree (10+ generations) sans text, with data problems like cycles and inbreeding?
I'm trying to write a program that will take genealogical data and represent a large amount of information at once. Currently the genealogical site I use can only display around 20 people on screen, with name, birthday, and birthplace. I want to create a visualization that is a higher overview, that allows quick navigation. The goal is to not show the name, birthday, birthdate, etc, but to show problems in the tree (missing information, impossible information (A is a parent of B, although A was born after B), cycles (A bore B, B bore C, C bore A), etc), and provide a link to inspect it in depth on the original site.
However, I have multiple problems. Theoretically, after 20 generations, I'm looking at 2^20 = about 1 million nodes on screen, However, after scanning though the data, it seems there is a high number of duplicate people, especially when people tie themselves into royal families. If there is a duplicate (cousin marriage results in duplicate grandparents), I'd only represent that node once. It turns out, ignoring duplications (which on average account for 25% of genealogical lines), this tends to average 50,000 nodes for the entire known tree of over 100 generations.
My idea was a 3d cone tree (pic related). Colors or shapes with a legend would be used to indicate problems/missing data. I'd attempt to show around 1,000 people at once (10 generations) with options to expand/collapse later generations. Other trees I've looked at using:
http://rp-www.cs.usyd.edu.au/~comp5048/Lect2-trees.pdf
TL;DR: how would you visualize an extend family tree (10+ generations) sans text, with data problems like cycles and inbreeding?
I'm trying to write a program that will take genealogical data and represent a large amount of information at once. Currently the genealogical site I use can only display around 20 people on screen, with name, birthday, and birthplace. I want to create a visualization that is a higher overview, that allows quick navigation. The goal is to not show the name, birthday, birthdate, etc, but to show problems in the tree (missing information, impossible information (A is a parent of B, although A was born after B), cycles (A bore B, B bore C, C bore A), etc), and provide a link to inspect it in depth on the original site.
However, I have multiple problems. Theoretically, after 20 generations, I'm looking at 2^20 = about 1 million nodes on screen, However, after scanning though the data, it seems there is a high number of duplicate people, especially when people tie themselves into royal families. If there is a duplicate (cousin marriage results in duplicate grandparents), I'd only represent that node once. It turns out, ignoring duplications (which on average account for 25% of genealogical lines), this tends to average 50,000 nodes for the entire known tree of over 100 generations.
My idea was a 3d cone tree (pic related). Colors or shapes with a legend would be used to indicate problems/missing data. I'd attempt to show around 1,000 people at once (10 generations) with options to expand/collapse later generations. Other trees I've looked at using:
http://rp-www.cs.usyd.edu.au/~comp5048/Lect2-trees.pdf
