Faculty of 1000? http://www.f1000biology.com/ - Martin Fenner

F1000 is very similar, but it's not a seedable recommendation engine. You're at the mercy of what is made available. I'm thinking of something more explicitly Pandora-like. - Mr. Gunn

So how is that different from pubmed's related articles? They're all indexed by humans (with machine assistance) - Christina Pikas

it's been around for a while and last time I checked, it wasn't free - Richard Akerman

Related articles is similar, and I don't actually know how they come up with the suggestions. I was thinking of something with a little more functionality, that could also be provided by a bookmarking service, to point out gaps in your reading. Last time I checked, "related articles" wasn't something you could get via eutils. - Mr. Gunn

http://www.biomedcentral.com/1471-21... <-- how pmra works - Joe Dunckley

Thanks, Joe! - Mr. Gunn from IM

(I actually glaze over at all the formulas and mathsy sounding words, so it didn't tell _me_ how it works. But I love that software/algorithm papers still always define precision, when even _I_ know that one!;) Do they think someone is going to get though the description of the algorithm, and then be stumped when precision is mentioned?) - Joe Dunckley

Having read it, it seems that it's a combination of word frequencies and MeSH terms(which are human supplied). I suppose they don't provide related articles via API because of the overhead involved? - Mr. Gunn

oh, and, re: f1000 -- i have no idea what their plans are, but haven't they hired rpg as their web2.0 guru, or something? perhaps they want to be more community driven? - Joe Dunckley

I think this is what the vision behind Mendeley is and yes I can only imagine that RPG as a force for development within F1000 will lead things in the same direction. To me the challenge is in finding the best way to leverage both the crowdsourced information and the expert information and to bring it together - Cameron Neylon

That's right, Cameron. To put things a different way, "What can we learn about the two approaches applied to scientific literature based on the differences in how last.fm and Pandora work for music recommendation?" Crowdsourcing recommendations themselves leads to Digg, but most scientists don't care what other people in other even fairly closely related fields find interesting. crowdsourcing the analysis/annotation, however, might get you the back-end data needed to really make relevant recommendations. - Mr. Gunn

*takes notes on how to do my new job* - Richard P Grant

I've mentioned to Victor/Mendeley that this would be a cool direction to go, and it's exactly the kind of thing a start-up may have the resources to tackle. Please keep us updated, Richard, on the developments at F1000. - Mr. Gunn

I don't think it's giving too much away to say that a version of 'crowdsourcing' is something we are thinking about seriously. How we do it (yes Richard: you have to pay for the reviews. That's the business model) without taking value away from our paying subscribers is going to be a matter for much discussion, I think... - Richard P Grant

Thanks, Mr. Gunn, for this great idea! I initially just wanted to comment here, then had a few more thoughts about it (still not fully baked), so I wrote a blog post instead: http://www.mendeley.com/blog... - Victor / Mendeley Team