Monday, July 09, 2007

Filtering papers on number of downloads

I was having a look at highly accessed papers for BMC Bioinformatics. In BMC, all journals have a page with the statistics of the most highly accessed papers of the last month. Several other journals now provide a similar service. The cool think about BMC is that they even tell you how many views per paper (sum of abstract, full text and PDF accesses on BioMed Central in the last 30 days). Not only that, the information in on the RSS feed they provide. That makes it very easy to feed into a pipe and have a threshold for number of views above which it will show up on the filtered feed.

Here is pipe example to filter out BMC Bioinformatic papers below 1000 views. The only problem is that the information is not stored as a number (example :"Number of accesses: 1226"). That is why I used a regular expression [1-9][0-9][0-9][0-9]$ instead of number filtering. I also don't know if the numbers are updated everyday .. but I hope so.

Even better would be to have some kind of service that given a DOI BMC would provide exactly this information structure. If other repositories provide a similar service then there is no point in worrying about the dilution in the number of page views because of open access because we could just sum views in the publishers site with Pubmed Central, etc.