Creating a 3D point cloud from an euclidian Distance Matrix

Question

I have a huge matrix that came out from a text similarity analysis like

elem/elem	text1	text2	text3	text4
text1	1	0.8	0.6	0.18
text2	0.8	1	0.73	0.29
text3	0.6	0.73	1	0.6
text4	0.18	0.29	0.6	1

I want to create a 3D cloud point that represent all my element in a 3D space with the distance between points according to the relative distance to all other elements

I would like to transform my table into something more like this:

points=[
{'text1':[x,y,z]},
{'text2':[x,y,z]},
{'text3':[x,y,z]},
{'text4':[x,y,z]},
]
edges=[
[[x,y,z],[x,y,z]],
[[x,y,z],[x,y,z]]
[[x,y,z],[x,y,z]]
[[x,y,z],[x,y,z]]
[[x,y,z],[x,y,z]]
]

I will implement the computation in python with numpy and pandas, and the rendering in a vueJS app, with a 3d lib such a D3js.

I'm for now searching for the right algorithmic approach to convert distance matrix into absolute 3D coordinate.

Thanks a lot for the help.

Answer 1

Thanks for the comment. i finally went for a force weigthed graph