At Wed, 16 Sep 2009 13:39:42 +0100, O.Stephens wrote: > > Thanks Erik, > > Yes - generally references to web sites require a 'route of access' > (i.e. URL) and 'date accessed' - because, of course, the content of > the website may change over time. > > Strictly you are right - if you are going to link to the resource it > should be to the version of the page that was available at the time > the author accessed it. This time aspect is something I'm thinking > about more as a result of the conversations on this thread. The > 'date accessed' seems like a good way of differentiating different > possible resolutions of a single URL. Unfortunately references don't > have a specified format for date, and they can be expressed in a > variety of ways - typically you'll see something like 'Accessed 14 > September 2009', but as far as I know it could be 'Accessed > 14/09/09' or I guess 'Accessed 09/14/09' etc. > > It is also true that the intent of a reference can vary - sometimes > the intent is to point at a website, and sometimes to point to the > content of a website at a moment in time (thinking loosely in FRBR > terms I guess you'd say that sometimes you want to reference the > work/expression, and sometimes the manifestation? - although I know > FRBR gets complicated when you look at digital representations, a > whole other discussion) > > To be honest, our project is not going to delve into this too much - > limited both by time (we finish in February) and practicalities (I > just don't think the library/institution is going to want to look at > snapshotting websites, or finding archived versions for each course > we run - I suspect it would be less effort to update the course to > use a more current reference in the cases this problem really > manifests itself). > > One of the other things I've come to realise is that although it is > nice to be able to access material that is referenced, the reference > primarily recognises the work of others, and puts your work into > context - access is only a secondary concern. It is perfectly > possible and OK to reference material that is not generally > available, as a reader I may not have access to certain material, > and over time material is destroyed so when referencing rare or > unique texts it may become absolutely impossible to access the > referenced source. > > I think for research publications there is a genuine and growing > issue - especially when we start to consider the practice of > referencing datasets which is just starting to become common > practice in scientific research. If the dataset grows over time, > will it be possible to see the version of the dataset used when > doing a specific piece of research? You might find the WebCite service [1] to be of some use. Of course it cannot work retroactively, so it is best if researchers use it in the first place. best, Erik Hetzner 1. http://www.webcitation.org/ ;; Erik Hetzner, California Digital Library ;; gnupg key id: 1024D/01DB07E3