From 26aec12147b12cc15b34e6b9c28e512ca8db67c1 Mon Sep 17 00:00:00 2001 From: David Marx Date: Sat, 13 Apr 2024 18:06:34 -0700 Subject: [PATCH] Update continuous_graph_embedding.md --- continuous_graph_embedding.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/continuous_graph_embedding.md b/continuous_graph_embedding.md index 34a0d8856..2f7c7e5b6 100644 --- a/continuous_graph_embedding.md +++ b/continuous_graph_embedding.md @@ -19,6 +19,13 @@ for non-trivial geometry (still assuming trivial topology, no holes) we can fit literal semantic relativity. QUAntiFIABLE semantic relativity. +what can we learn by analyzing that shell? +- volume: how "big" an idea is..? important? abstract (i.e. hierarchies would nest into matryoshka subsets)? +- smoothness: how uniformly related it is to its neighbors? no idea.. +- gaussian curvature: some kind of local "density" or "mass". how tightly concentrated the local neighborhood is. +- sharpness/fuzziness, rate of transition over boundary, permissivity at boundary: how strictly defined the entity is in the space + +oooo.... I should do a geometry of semantic space survey ## Expected Properties