Last week, Bloggers for Peer-Reviewed Research Reporting announced a post-aggregation system for posts discussing peer-reviewed research only. They didn’t give any details of exactly how the aggregation system would work, so I bugged the people behind it via email.
Dave Munger is in charge of the project, but Zachary Tong is the one in charge of developing the aggregation system itself. Here’s the email I received from Zachary (reprinted with permission):
- Bloggers register for an account at BPR3.
- We require bloggers to include a small snippet of code in every post that should be indexed by BPR3. This snippet of code contains the citation data for the journal in question and uses the open COinS format (I noticed on your blog that you are familiar with it, so I’ll spare an explanation).
- As a tool to the blogger, we provide a “citation generator” that can be used to create the required COinS tag as well as include the icon or plaintext citation (if desired).
- The indexer at BPR3 routinely checks the RSS feeds of registered blogs. When the indexer notices a new post, the permalink is loaded and scanned for a COinS tag. If the COinS tag is found, it is parsed into the database and displayed on the frontpage of BPR3 (assuming the blog in question has been activated by a moderator).
As I’m sure you’re aware, the use of the COinS format allows multiple vendors and systems to make use of the citation data, as well as providing a fully developed format for our use. No need to reinvent the wheel 🙂
It also allows blogs to invisibly include citation data (and be indexed by BPR3) without displaying the icon, which some people have expressed interest in.
Note that the thing which causes a post to be added to the aggregation system is the presence of COinS metadata, not the icon, as you might assume from all the talk at scienceblogs.com about adding the icon to posts. The icon is simply a way of advertising the project and indicating to the reader that the post is part of bpr3.org’s system.
This mostly addressed the concerns I had, and I’m glad to see that the developers have an understanding of current metadata-based approaches for identifying citations in a web page. Without metadata, their system wouldn’t have any advantage over a technorati search, but with the journal-specific metadata included, this data can be collected and reused in ways hard to predict at the moment. (Did the del.icio.us team predict this?) The use of RSS to detect new posts is also good, because it’s a well-established technology for identifying new content without consuming a lot of bandwidth. Another thing I was glad to see was that the developers understand the need for an easy-to-use metadata generator. Tagging systems, like those used by del.icio.us, Connotea, and WordPress are an example of a successful implementation of metadata, requiring little effort at the time of publishing and providing a substantial benefit for subsequent users of the system. It seems to me that achieving “effort parity” with linking to the abstract in pubmed is the critical thing needed for widespread adoption of bpr3.org’s system. Unfortunately, the prototype of the citation generator, which looks rather like the COinS generator provided by Worldcat, still calls for quite a bit of manual entry of citation information, as opposed to looking it up from the PMID or DOI, but I feel certain that this needed functionality will be coming in the finished product.
I think this is an exciting project and I will be following it closely.
I have been traveling, hence the delayed response to your comment on “Your Cheatin’ Heart” at http://blogs.nature.com/nm/spoonful/2007/09/your_cheatin_heart.html.
Chris Muller
That’s OK, Chris. It’s actually Juan Carlos that I was addressing my comment to. I find it hard to believe that he would suggest that academic dishonesty is extremely widespread, hint that he has data to back that view up, and then not expect someone to ask him to show that data.
I don’t disagree at all with your contention that some PIs would rather believe that their students messed up than accept that it might be their theory that’s wrong. I do disagree that it’s an extremely common occurrence, and I think in his post he was maybe a little guilty of what he’s accusing people of.
Mr. Gunn,
First, thanks for the critical evaluation. We are striving to stay a community-based project and critical comments like yours and others helps keep the project moving in the right direction.
I just wanted to say one little thing about DOIs. We are currently in the process of getting CrossRef access to DOI metadata. Unfortunately, because we aren’t a library organization, we fall under “Affiliates” and are subject to fees and whatnot. We are currently negotiating with them about this, but as with all things involving corporations, it is going to take a while.
We agree that PMID/DOI references are the way to make this system both powerful and easy to use. Until we can get the license though we are a bit hamstrung. Which is why we rolled out the manual citation generator for now, to keep momentum in the project.
I’m interested in all this because I’m a scientist who both blogs and publishes in peer-reviewed journals. Clearly, we need a good way to reference peer-reviewed research within a post and we need a good way to reference posts within peer-reviewed research. More than that, though, we need the people who have the experience, good ideas, technical know-how, and the ear of eager users to get together and make something that will not only be widely used, but work well too.
Alf and colleagues have worked on this for some years now, so it’s not surprising that they’ve reacted somewhat negatively, but since they’re willing to work with you and they’ve apparently got the metadata discovery issue worked out, why not join forces?