| Hi Yves
Greenstone doesn't have very good support for multilingual metadata at
the moment.
Here are a couple of suggestions to achieve what you want:
1. have three different metadata elements, one per language. eg
dc.SubjectFrench, dc.SubjectEnglish, dc.SubjectGerman. Assign each
metadata to each document, then build separate classifiers on each one
(classify Hierarchy -metadata dc.SubjectFrench, etc.).
2. Create three hierarchy files, one for each language (see Section 2.2
in developers guide for format).
Each line is like
id position value
Create the three files the same, but with different values. then you
only need to assign the id as dc.Subject metadata (language independent)
e.g. in dc.Subject.en.txt
1 1 Animals
1.1 1.1 Duck
...
and dc.Subject.fr.txt
1 1 Animaux
1.1 1.1 Canard
...
In this case the id is the same as the position, so for doc1, the
dc.Subject metadata will be 1.1. You only need to add one piece of
metadata for each document.
Build Hierarchy classifiers on each file
(classify Hierarchy -metadata dc.Subject -hfile dc.Subject.en.txt, etc.)
Obviously the second method is better if you have a controlled vocab and
you know what it will be at the start.
By default, both these methods will result in three subject buttons. You
will need to modify the navigation bar to link to the different
classifiers based on the interface language.
We can help you do this when you get to that stage.
I hope this helps,
Regards,
Katherine
Yves Deplasse wrote:
> Hi,
>
> I'm trying to build a pdf collection. The documents are rather
> technical ; some of them are written in english, others in french or
> german.
>
> The people who will assign metadata to each document know one
> language only (i.e. french) so the value for -say- the dc.subject
> metadata will be written in french.
>
> My goal is : when a french-speaking user navigates the collection and
> browses by subject, the original (french) values are displayed AND
> when an english-speaking user browses the same documents, english
> values are displayed.
>
> Since the subject values might fit in a controlled vocabulary, does
> anyone think i might use some kind of dictionary in order to reach my
> goal ? I would like to avoid duplicating the metadata over and over.
>
> Browsing will rely on AZlist and Hierarchy.
>
> A quick example ?
>
> When setting 'Preferences' to English, a user would be able to browse
> by subject and get :
>
> Subjects
>
> Animals
> -Duck
> --> doc1
> --> doc 2
> --> doc 3
> -Mouse
> --> doc 4
>
> Setting preferences to French :
>
> Sujets
>
> Animaux
> -Canard
> --> doc1
> --> doc 2
> --> doc 3
> -Souris
> --> doc4
>
> Canard is french for duck, souris = french for mouse.
>
>
> Same documents, differents labels
>
>
> Any idea ?
>
>
> --Yves Deplasse
> Brussels, Belgium.
>
>
>
>
>
>
>
>
> _______________________________________________
> greenstone-users mailing list
> greenstone-users@list.scms.waikato.ac.nz
> https://list.scms.waikato.ac.nz/mailman/listinfo/greenstone-users
> |