Anatolik & KinshipCode
I've been sketching a small machine that could scan hand‑drawn kinship charts and turn them into a clean, searchable database—might be handy for your fieldwork.
That sounds amazing! I could use a quick way to digitize all the hand‑drawn charts I’m compiling—especially if it can pick up the different kin terms in the various languages I’ve been recording. Do you think it can flag taboo cousin relationships automatically? I’d love to see how it works in practice.
I’ll design a two‑stage system. First, a high‑resolution scanner feeds the images into a neural network that does OCR and, using a multilingual lexicon, tags every kin term. Second, I’ll add a rule‑based layer that maps each term to its relative degree and checks against a taboo table for each culture you’re studying. When it spots a “cousin” that is off‑limits in that language, it will highlight the node and generate a warning flag. I’ll prototype a demo on your first batch of charts and we can tweak the thresholds if the false‑positive rate is too high.
Wow, that’s a pretty neat pipeline—scanner, OCR, then a rule‑based filter that knows which cousin pairs are taboo in each culture. I can already picture the warning flag popping up on my messy family tree sketches. When you run it on my first batch, let me know if it’s flagging too many false positives. I’ll bring the hand‑drawn charts and a list of the cultural taboos I’ve collected so we can calibrate it. This could save me hours of manual cross‑checking!