[open-bibliography] FRBR examples

Thu May 27 17:12:29 UTC 2010

I don't know enough about FRBR to comment on its strengths and
weaknesses in detail, but I'm concerned that people seem to be
confusing a data model with the user visible manifestation (not
Manifestation) of that model.  There's no law that says you have to
have one data entry/display screen per entity.  Hopefully none of your
current system make visible how many database tables were used in
their construction.  A software engineer is free to put together a
"book" page that's got a little Manifestation data with a sprinkling
of Expression info and a taste of user specific data if that's what's
useful for people.

There's a lot of power and value in having a rich data model, but that
needs to be tempered by the need to manage complexity to the bounds
which are understandable and implementable by mere mortals.  It's
disappointing though to see people arbitrarily saying things "aren't
important" because they have little value in their world view.  One
reader may want a story in any language they read, but a different
scholar may be interested in comparing different translations of a
work in the same language or across languages.  The model needs to be
able to store the information to make both possible.  The simple stuff
should be easy, but the more complex should still be possible.

User models have limited transferability to computers because users
are so much better at dealing with ambiguity.  I can borrow a book
from my library (meaning the library system as represented by the
bookmobile that visits), go to a reading at the library (meaning the
local neighborhood branch library), or go to a concert at the library
(meaning the old library building which now serves as community center
and which has a different official name, but everyone still calls the
"Hunt Library") and anyone with the same shared context can easily
disambiguate all the different "libraries."  Computers aren't that
smart and these three different kinds of things all need to be modeled
differently.

I'd echo Tim's sentiment that any practical modeling effort needs to
be informed by real world experience.  You'll still need to separate
out whether people are complaining about lousy UIs vs unusable data
models, but there's no substitute for real world data and live users.
I had a quick look at the LibraryThing site and didn't see anything
about their data model and the license looked pretty scary ("We have
yet to settle on a license. You agree to abide by whatever license we
eventually choose." - Yikes!!) so I didn't want to poke too far, but
if they were willing to share, it would be a good data point.

The Freebase publishing domain (http://www.freebase.com/view/book)
only has 3 million books, but it's got a relatively rich data model
[1] which supports works, adaptations[3,4], and translations[5,6,7].
Another strength is that it isn't limited to the publishing domain, so
one can go from a book to the play to the performance of the play to
the movie that was made from the play to the soundtrack album for the
movie or from your naval architecture book to the ships the author, a
naval architect, designed, etc, etc.  I'd encourage folks to think
about how book related things are connected to the rest of the world.
Books are cool, but they're just a tiny slice of the world.  I think
collaborative schema development is another Freebase strength.  Anyone
can extend the schema, propose the extensions for promotion to the
commons, or discuss the existing schema and it's all done in situ with
live data for people to work with.  This allows incremental
improvement of the schema over time (refactoring the data as you go).

The schema for translations currently allows translations to either be
simply entered as another edition of a book, in relatively
undifferentiated, but simple, fashion or, with some extra effort, to
enter them as a separate work, linked to the original, with
translator's name, etc.  Freebase commits the error of having the UI
tied tightly to the schema that I said to avoid and because of this I
think the translation schema is at (or beyond) the limit of complexity
that a user can deal with.  Being able to experiment with actually
entering data provides much more insight into this than studying a
specification would.

Sorry this is so long, but one final question in conclusion.  What's
the point of this discussion?  I suspect the FRBR committee is off in
standards committee mode and will do whatever "professional library
scientists" deem is appropriate.  Is the goal to try to get them to
change or to propose/implement an alternative or just kvetch about the
general state of bibliographic schemas?

Tom

[1] http://www.freebase.com/type/schema/book/book
[2] http://wiki.freebase.com/wiki/Entering_Data_for_a_Book
[3] http://www.freebase.com/view/topic/en/adaptation_2002
[4] http://www.sandbox-freebase.com/view/user/tfmorris/default_domain/views/adapted_adaptations_topics
[5] http://wiki.freebase.com/wiki/Entering_a_Translated_Work_of_Literature
[6] http://www.freebase.com/edit/topic/guid/9202a8c04000641f8000000005136572
[7] http://www.freebase.com/view/m/04dg18

On Thu, May 27, 2010 at 10:51 AM, William Waites
<william.waites at okfn.org> wrote:
> On 10-05-27 15:40, Christopher Gutteridge wrote:
>>
>> Such relationships should be described by some kind of extension, if
>> at all. Describing relationships like this semantically is difficult
>> and of limited value for the effort and complexity required.
>>
>
> I strongly disagree about the "limited value" part. If one is concerned
> about running a
> library that might be true. If one is concerned with mapping
> researchers' knowledge
> about works and writers in their fields then it goes right to the core
> of the project.
>
> It is still hard though
>
>> If I want to read a copy of the Lion the Witch and the Wardrobe, then
>> any copy in a language I speak is fine. An audio book could/should be
>> considered as just-another-manifestation, but a film or audio drama is
>> just a related work. Relationships between different works is an
>> entire rabbit hole of its own which is not essential to providing a
>> useful model. It can be added to a working model later, probably with
>> different datasets requiring different ways of modelling such things.
>
> This goes back to my original question. Rephrased, is FRBR a suitable
> working model
> for people to hang their more elaborate statements off of? Or is it
> confusing enough that
> people won't be able to consistently identify which subjects and objects
> they mean?
>
> My feeling at this point is to go with a much flattened
> Work-Manifestation model and
> freely create new Works where needed (e.g. translations) with a
> predicate like
> derivedFrom or a sub-predicate like translationOf as needed.
>
> Cheers,
> -w
>
> --
> William Waites           <william.waites at okfn.org>
> Mob: +44 789 798 9965    Open Knowledge Foundation
> Fax: +44 131 464 4948                Edinburgh, UK
>
> _______________________________________________
> open-bibliography mailing list
> open-bibliography at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/open-bibliography
>