On 03/09/2008 17:03, "Peter Millington" <[email protected]> wrote:
While author's final post-refereed draft is sufficient and acceptable for open access and research purposes, it is not the best. The best is the published version (publisher's PDF if you will). But it definitely isn't the best for text-mining purposes, which is where we're aiming for, isn't it? The technology and the knowledge of the future and all that. Step aside from the world of publishing and citing, remember what science is all about and take the big view. The quotes (given in 2006) from three leading experts on the technicalities of text-mining, when I asked them whether PDF is an acceptable (and usable) option, or whether we should be continuing to encourage authors to deposit their own, final version in some other format (Word, LaTeX, etc): John Wilbanks, Science Commons (when asked about the possibility of `scraping' PDFs for text-mining: "Scraping is the right word, because having to work with PDF is really scraping the bottom of the barrel." Peter Murray-Rust, Cambridge: "Getting to XML from PDF is like starting with the burger and trying to get back to the cow." Cliff Lynch, CNI: "PDF is evil." If an article has been published by the time someone wishes to cite it, then they cite the published article with all its page numbers etc (plus the URL of the repository version so that everyone can read the thing). If people wish to cite an article where there is not yet a definitively published version, but they have miraculously found an open access postprint, they can always adhere to age-old practice and say `Journal of XXX, in press'. When people did that in the `old days' (my era) it was a pain, because locating the article really meant waiting for BIOSIS or Index Medicus to come out in print, or trekking off to another university's library, if your own library didn't subscribe to the journal. Nowadays, locating it online from an `in press' citation is a doddle. Alma Swan Key Perspectives Ltd Truro, UK
