So this is my recommended framework for proceeding. Tim, I'm afraid
you'll actually have to do the hard work yourself.
No, I don't. Because the work isn't fundamentally that hard. A
complex standard might be, but I never for a moment considered
anything like that. We have *512 bytes*, and it needs to be usable by
anyone. Library technology is usually fatally over-engineered, but
this is a case where that approach isn't even possible.
Jonathan did a very well summary - you just have to pick what you main
focus of embedding bibliographic data is.
A) I favour using the CSL-Record format which I summarized at
because I had in mind that people want to have a nice looking citation
of the publication that someone tweeted about. The drawback is that CSL
is less adopted and will not always fit in 512 bytes
B) If you main focus is to link Tweets about the same publication (and
other stuff about this publication) than you must embed identifiers.
LibraryThing is mainly based on two identifiers
1) ISBN to identify editions
2) LT work ids to identify works
I wonder why LT work ids have not picked up more although you thankfully
provide a full mapping to ISBN at
http://www.librarything.com/feeds/thingISBN.xml.gz but nevermind. I
thought that some LT records also contain other identifiers such as OCLC
number, LOC number etc. but maybe I am wrong. The best way to specify
identifiers is to use an URI (all relevant identifiers that I know have
an URI form). For ISBN it is
For LT Work-ID you can use the URL with your .com top level domain:
That would fit for tweets about books with an ISBN and for tweets about
a work which will make 99.9% of tweets from LT about single publications
C) If your focus is to let people search for a publication in libraries
than and to copy bibliographic data in reference management software
then COinS is a way to go. COinS is based on OpenURL which I and others
ranted about because it is a crapy library standard like MARC. But
unlike other metadata formats COinS usually fits in less then 512 bytes.
Furthermore you may have to deal with it for LibraryThing for libraries
Although I strongly favour CSL as a practising library scientist and
developer I must admit that for LibraryThing the best way is to embed
identifiers (ISBN and LT Work-ID) and maybe COinS. As long as
LibraryThing does not open up to more complex publications like
preprints of proceeding-articles in series etc. but mainly deals with
books and works this will make LibraryThing users happy.
Then, three years from now, we can all conference-tweet about a CIL
talk, about all the cool ways libraries are using Twitter, and how
it's such a shame that the annotations standard wasn't designed with
libraries in mind.
How about a bet instead of voting. In three years will there be:
a) No relevant Twitter annotations anyway
b) Twitter annotations but not used much for bibliographic data
c) A rich variety of incompatible bibliographic annotation standards
d) Semantic Web will have solved every problem anyway
Jakob Voß <jakob.v...@gbv.de>, skype: nichtich
Verbundzentrale des GBV (VZG) / Common Library Network
Platz der Goettinger Sieben 1, 37073 Göttingen, Germany
+49 (0)551 39-10242, http://www.gbv.de