Finding Conversations around Academic Publications

In Jon Udell's post Bloggers talk to bloggers, scientists talk to scientists earlier this week, he and others in the comments wondered if it was possible to
  1. normalise all the identifiers - URLs, PMIDs and DOIs - relating to papers, and
  2. collect together the conversations around a paper from all the various academic, news, blog and other discussion channels.

And it is: with the APIs from ask.connotea, Connotea, Postgenomic, PubMed, PubMed Central, Scopus and CrossRef you can pull together a fair amount of information relating to discussions, bookmarks and citations of a paper identified by either URL, PMID or DOI.

Here's the form to enter a URL, and here's the discussion for the paper that Jon used as an example, identified by URL, DOI or PMID.

(It's having to call quite a few APIs sequentially, so give it a little while to return the results).

-------

TODO: Pull in comments attached to papers on publishers' sites; use Lucene indexes of PubMed to match press releases to published papers; show more of the PubMed and Postgenomic conversation and citation graphs; fetch shared annotations from Zotero.

-------

Update: Added Bloglines and del.icio.us APIs, but those won't work so well because they don't normalise the URIs.

January 24, 2008

Comments

Great ! This could be use to correlate different measures of impact :)

Posted by: Pedro Beltrao on January 25, 2008 2:31 AM

Great idea. Unfortunately the form does not load form me, is it still alive? Can you publish some source code?

Posted by: Jakob on January 25, 2008 11:22 AM

Argh. There's a problem with the firewall in front of Scintilla - it doesn't like Ubuntu (hopefully it'll get fixed soon).

Posted by: alf on January 25, 2008 1:02 PM

Any chance of running this as a webservice? If you could just pass back the results in JSON format (with an arbitary but defined structure), it would be very easy to annotate DOIs on journal pages.

Posted by: baoilleach on February 1, 2008 7:09 PM

Unofficially, try putting &format=json on the end.

Posted by: alf on February 4, 2008 4:51 PM

That is looking very cool :-) Everything except "ciations.pmc" and "metadata.crossref.query" seems to be working fine. (Tested with the DOI link in your post above)

Posted by: Noel O'Boyle on February 4, 2008 5:08 PM

This is looking real nice, Alf.

Posted by: mr. gunn on February 7, 2008 10:14 PM

Thanks for the json output!

I plan to incorporate this into my Umlaut link resolver next week probably. Very cool.

Not sure what to do with 'url'. I know what to do with doi and pmid, but in most cases the URL for an article I will have will be vendor-specific and locked behind a for-fee door. Not sure if it's worthwhile to even bother trying to pass that URL to your service. Thinking not. Any ideas though?

Posted by: Jonathan Rochkind on August 22, 2008 10:59 PM

Oops, I see the json response might not be there yet after all.

Can I get your email address to talk to you my intended use more over email?

Very exciting little tool.

Posted by: Jonathan Rochkind on August 22, 2008 11:03 PM

A bit late coming to the party :(

Have you done this in php? I'm wondering how the parsing of PubMed xml is done...

Posted by: Tom Pasley on November 2, 2009 11:21 PM

All fields are optional, email address will not be shown; no HTML, URLs are automatically hyperlinked.