Gisle Hannemyr wrote:
> But I make the provision that the human user only sees the data
> indirectly (i.e. mediated by the computer), and that the computer
> should only mediate what it believes make sense. This provision
> filters out all relations "enriched" with pseudo-natural-language
> qualifiers that the computer has not been explicitly programmed to
> recognise.
No argument from me.
> > I don't feel that we should worry about present limitations of the
> > search engines. If we want to label the fields separately, we should do
> > so; and the technicians should deal with it. If we don't want to label
> > the fields separately, so be it.
> > MARC can handle these distinctions, and that system is over 30 years
> > old.
>
> Well, as a technician, I worry (a lot)!
>
> MARC has created and specified a very detailed framework to handle
> these distinctions, based upon a stringent and detailed coding system.
>
> The DCI has not. It has created a set consisting of 15 core properties
> (this is good) and a syntax that allow cataloguers to express all sort
> of complex relations (MARC-like and non-MARC-like) in pseudo-natural
> language (this, IMHO, is bad). Making computers understand this sort of
> pseudo-natural language is very difficult -- much more difficult than
> interpreting a completely specified coding system like MARC).
I agree with some of this. From another point of view, you get a lot
less argument using MARC-like numbers. You don't hear many people
yelling at each other: "I think it should be 100!" "I think it should be
101!"
I'm a MARC cataloger, and as such, the coding is of little importance to
me. It's how the coding will be used, and the information that will go
into the coding that is my area.
Still, I must disagree. The limits of the system should be determined by
the designers, the technicians should be there to build a system that
fits the design. The design should not be changed because the system may
be more difficult to build at the moment.
Of course, there are always negotiations between those who design and
those who build, and maybe we are stuck with an English-language set of
coding. But unless we want to rewrite DC completely, (something I don't
advocate at all) we're stuck.
> Most of the people that will put metadata on the web will not have any
> type of formal training as cataloguers. They will take whatever tools
> and guidance we provide (hopefully) and some of them will try to do a
> job faithful to the spirit of the DCI. Others will use the same tools
> to create index spam.
I agree--and unless we consider it now, it will be the spam that wins.
Good, valuable material won't stand a chance.
But that's another issue.
Jim
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|