Open JarbasAI opened 7 years ago
current parsing gives following info from search term
description: English comedian, actor, writer, presenter, and activist
what: human
data: website : http://www.stephenfry.com category : Category:Stephen Fry citizenship : United Kingdom image : Stephen Fry cropped.jpg instance : human IMDB : nm0000410 birth : +1957-08-24T00:00:00Z movement : atheism
properties: P135 : [u'Q7066'] P345 : [u'nm0000410'] P910 : [u'Q8817795'] P27 : [u'Q145'] P856 : [u'http://www.stephenfry.com'] P569 : [u'+1957-08-24T00:00:00Z'] P18 : [u'Stephen Fry cropped.jpg'] P31 : [u'Q5']
property lookup and parsing would give meaningful connections, from description distant parents can be extracted from what field close parent can be extracted, data field parsing is harder and maybe should be considered info and not get connections from there
currently scraps following node info:
should parse properties for connections
original issue:
a backend should be made to adquire knowledge from wikidata #4
wikidata connects subjects in useful ways, we should keep an eye on properties: https://tools.wmflabs.org/hay/propbrowse/
we should map the correct properties to either childs or parents, or even as arbitrary data connections #10
this seems the most well structured source for our purpose, surfing github for stuff of this kind i found this promising start -> https://github.com/Dorialexander/Wikidaesthetics