The Reddit world map – How to map out online social networks | Hacker News Reader

The Reddit world map – How to map out online social networks

randalolson.com

74 pointsrhiever12 comments

I'd really like to see this kind of map for scientific publications. That is, instead of users linked by likes and such, a map of publications linked by citations. It seems this would make it easy to find the big papers in a given subject. I've looked and it seems like there are a few attempts at this out there, for example by IEEE, and Reuters (behind a paywall?), but the ones I've seen don't quite get it right. Maybe I've missed it, or maybe it's more difficult to do than it sounds?

A world map of science publications. I think it could be an awesome visualization, but more importantly, it would be really useful.

That's a great idea! I've heard of researchers doing this kind of analysis to detect "hidden gems," i.e., papers that influenced papers that ended up becoming the big cited paper. The tough part, of course, is that most journals are keen on keeping all of their data behind company walls. Perhaps PLoS would be an exception, though...

I've used such a tool at university but I've now found out it's a subscription service. You could get graphs of papers that were cited and could use it to find papers that were roots of a certain discipline. Though it wasn't as advanced a graph analysis as you propose it was pretty helpful.

And you're right about journals sitting on that citation data. The service I used was

http://en.wikipedia.org/wiki/Web_of_Science

Microsoft Research has a prototype that is headed in that direction:

http://academic.research.microsoft.com/PaperCitationGraph#19...

It requires Silverlight. It also isn't showing a full map (but the idea is sort of there)

http://academic.research.microsoft.com has some other tools for looking at the connections between papers and authors, and some of it does not require Silverlight.

I think just getting the data is the hardest part; there are several corpora available in academia, but not so much outside it. Although you wouldn't need the full text of the paper, just the bibliography entry, which could be much easier to get. (My PhD research was initially on scientific publication citations but the corpus we had was small, ~200 documents.)

Scientific papers are also interesting because you can use co-author relationships as a secondary graph. You could probably infer all kinds of things from that!

I would like a world map of philosophy publications - both articles from journals and works throughout the ages.

Instead of mapping by citation I would like this world map to have its geographic boundaries demarcated by topic (I think).

You see, philosophers don't even agree what constitutes the sub-disciplines of their discipline so I figure we could find out empirically by mapping.

Of course this presupposes that we can identify papers and works as firmly philosophical in the first place when there are bound to be edge-cases.

I realise that my idea may be a bit fuzzy but I accept that it is a proto-idea, not you fully formed. Thoughts?

I actually spent a few months putting one together a few years ago using PLOS, and it was actually fun.

Unfortunately outside of PLOS and arXiv, the APIs for building the graphs just don't exist.

Here's one of the better graphs I generated: https://github.com/drewbuschhorn/DoctorMoon/raw/gh-pages/Scr...

https://github.com/drewbuschhorn/DoctorMoon

which is a graph centered from Why Most Published Research Findings Are False - PLoS - Ioannidis JPA

Which was a great center case to work from since his name is misspelled often in the literature, and he's done lots of work in PLOS.

Never could get anyone interested in it, but email me at drewbuschhorn @ gmail.com if anyone has any questions.

The Smithsonian/NASA Astrophysics Data System[0] has an API[1], though I am not sure it has all the information one would need to create such a graph.

[0] http://adsabs.harvard.edu/

[1] https://github.com/adsabs/adsabs-dev-api

Thanks! I had seen the main site, which I had used lightly in analyzing arxiv papers, but I'd completely missed that they had a library on github.

Someone recently did something like this for a programming language research conference. http://www.pl-enthusiast.net/2014/10/21/is-popl-one-communit...

MIT Immersion does something very similar for your email. https://immersion.media.mit.edu/

Sounds cool. Unfortunately their video did not play for me.

Reply on news.ycombinator.com