[extropy-chat] Intellectual influence

[Anders Sandberg]

Neil H. wrote:
>
> Interesting stuff. A few things I wonder about:
>
> * If you use the "influences" and "influenced by" information in the
> Wikipedia templates, does it have similar results?

I didn't know the Big W had that feature, and now I'm itching to try. Just
simple pattern matching in a python script and I could make a crawler to
harvest the graph. I just have to figure out the best way of download a
copy of the Wikipedia to play with locally. Anybody who has done it?

> * How similar are the results if you simply rely on which person articles
> have wiki links to each other?

My guess is that the influence links will have a higher signal-to-noise
ratio, especially since many person references relate more to interactions
and family than intellectual links.

It also depends on how well you can disambiguate persons from other
articles. I didn't see any "person" category, which means it is likely
that it will be tricky to automatically determine whether a link is
another person (I'd love to have a good named entity reference detector
for English to integrate in my software - in Swedish capitalisation can do
a surprisingly decent job, but it fails for English). Even if we just
accept philosopher category pages it will be somewhat messy. Still, in the
long run every link is sacred: it contains a deliberate relation to
another document we can mine if we can just dig out enough context around
it.

Why am I looking at this problem in the first place? Let's just say the
Future of Humanity Institute has a few empty walls screaming for beautiful
visualisations of the world and our understanding of it.


-- 
Anders Sandberg,
Oxford Uehiro Centre for Practical Ethics
Philosophy Faculty of Oxford University