[greenstone-users] Length of a classifier list

From Jonathan Gorman
DateFri, 2 Dec 2005 08:47:07 -0600 (Central Standard Time)
Subject [greenstone-users] Length of a classifier list
Hello again,

I feel a little foolish asking this question, but I'm afraid of spending a
lot of time looking at the wrong places. Is there a way to have some
predictability as to the number of items shown on a page within a
browse. I know there is min and max grouping, but what I would like is
to be able to say that there will be always 10 items per browse. I'm
pretty sure I've seen this somewhere. I guess I'd prefer not having to
control it but rather be able to predict how the classifier will be set
up.

Why I'm asking:

My end goal would be to have links contained within a generated html
document for important terms. So a paragraph in a recipe has "and wait
the meat is browned". I'd like to link that to the page in the "cooking
terms" classifier where "brown" appears. There seems no nice way to do
this in Greenstone so here is my worse case scenario:

I have XML documents so it seems not to difficult to have something like
<cooking_term index="brown">browned</cooking_term>. Then when I convert
it to the HTML I have Greenstone process, I can replace that with a link
by looking up the attribute "index" in an index (not Greenstone-based)
that contains the terms and where they will appear (ie CL3.2.3). Then I
just have to make the link <a
href="http://url.edu/cgi-bin/library?a=d&c=recipes&CL=3.2.3">browned</a>.

The trick is how to generate that index. Worse case scenario I go ahead
and generate the Greenstone library. Then I can spider over it and
determine what pages the terms occur on in the classifier.

Basic algorithm would be to do a depth search, starting at CL1, seeing if
there are items, if so try CL1.1 and keep going until there is no items,
then go back up one level in the hierarchy and check the next sibling.

The other approach is to define my own hierarchy but I would like to avoid
that if possible since some of the terms I'd be adding probably wouldn't
be very well controlled.


Jonathan T. Gorman