Re: [greenstone-users] questions about a configuration: collection of art periodical images

From Federica Zanardini
DateTue, 24 Feb 2004 11:14:50 +0100
Subject Re: [greenstone-users] questions about a configuration: collection of art periodical images
In-Reply-To (20040217-185855-607961619-tao-lib-uchicago-edu)
Thanks Tod,
I solved the problem using the hierarchical classifier with a table containing the hierarchy Title-Year-Issue.
This create a list like the subjects tree where the leaf are the issues.
Then for each issue I display n-pages that contains the thumbnails of ten pages scannered at time in a table and for each thumb, in the right, some metadata and the link to the high resolution image.

Now I have an other question to submit to the list:
are there some hardware requirements that greenstone need in order to manage approximately 100.000 documents?
A Intel server with Xeon Processor 3.2 GHz/533 MHz  and RAM 1 GB can go well?


At 18.58 17/02/2004 -0600, Tod Olson wrote:
>>>>> "FZ" == Federica Zanardini <> writes:

FZ> I have installed successfully Greenstone on Debian, and now I am
FZ> trying to configure the library, but I need of some suggestions.
FZ> Mine is a collection of art periodicals (about 200 titles ) to
FZ> everyone of which corresponds from 1 to n years (and 1 to n issues
FZ> for every year). For each issue we have scanned every pages (one
FZ> image file for each page).

FZ> I'd like create the following lists to browse, access and display
FZ> the serials:

FZ> 1. List of titles
FZ> 2. (after the selection of a title) date list of years for this
FZ>    serial
FZ> 3. (after the selection of the title and year)  date list of
FZ>    Issues

You mean something like this?:

Journal of Art Today
                no. 1
                no. 2
                no. 3

That's something the Hierarchy classifier can do.  I think the
Hierarchy classifier will only work with one Metadata field.  If so
you would have to arrange a special field where title, year, and issue
are all together.  Could probably write a program to build this from
the separate metadata.

I have some more specific questions and desires regarding the
Hierarchy classifier that will go into separate email, later.

FZ> 4. (after the selection of the Issue) display of the page-images using a
FZ> slide-show
FZ> (like this:
FZ> )

That is a very customized use of Greenstone. The page turning/slide
show thing is entirely custom-generated from METS documents.  We can
talk about how it works, but it would require some programming
resources to implement at your site.

There is a default page turner available within Greenstone.  If your
documents are organized into Sections where the Title Metadata is
numbers, you should get the default page turner.  Something like
(closing tags omitted):

                <Metadata name="Title" value="JoAT, 2000, no.1">
                        <Metadata name="Title" value="1">
                <Content>&lt;img src="page1.jpg"&gt;
                        <Metadata name="Title" value="2">
                <Content>&lt;img src="page2.jpg"&gt;

That's how I remember it, anyhow.

FZ> Sorry, I'm newbie and I have many boring questions: I have not found
FZ> between the examples a similar one to what I want to obtain.

FZ> 1. which it is the better method in order to organize the files
FZ> and in order to associate them the metadata ?  I have thought in
FZ> directories organized to tree (collection --> periodical --> Year
FZ> --> issue --> images of pages) and in each level (directory) I
FZ> store a metadata.xml file. But it seems to me that nearly no
FZ> digital library present in the Greenstone website as example has
FZ> the metadata in a separate file. It is perhaps better to create
FZ> HTML pages in order to contain only the metadata?

FZ> 2. I have not still well understood like constructing the indexes
FZ> and as to show them: for example to build the title list (first
FZ> level of browsing), it is better to use a hierarchy classifier
FZ> instead the AZlist? In fact my list of serials titles is fixed an
FZ> I can create a list of them (like a subject list).

I suggest that you build a few HTML files by hand that you can
experiment with.  Put some images for an issue into a file, then add
the <Section> tagging in comments as described in the documentation.
Do this for a few issues.  Then you can see what the default page
turner looks like, and experiment with some of the classifiers. 

Tod A. Olson <>     "How do you know I'm mad?" said Alice.
Sr. Programmer / Analyst            "If you weren't mad, you wouldn't have
The University of Chicago Library    come here," said the Cat.

Federica Zanardini

Servizio Banche Dati in Rete di Ateneo
Div. Coordinamento Biblioteche
Universita' degli Studi di Milano
Via G.Colombo,46 - 20133 Milano - Italy