Re: [greenstone-users] Metadata in MGPP

From Katherine Don
DateMon, 11 Aug 2003 14:49:27 +1200
Subject Re: [greenstone-users] Metadata in MGPP
In-Reply-To (Sea1-F65W8nADeLXMyc0000a0c8-hotmail-com)
hi dave

> Hi,
> I just have a question concerning about Metadata. If I assign a subject
> metadata to a document in metadata.xml. Will this subject metadata be
> applied to only the document as a whole or to individual sections within the
> documents too. Because I realzie that the date metadata which I assign to
> the document doesnt seems to apply to the sections nested inside the
> document.

it only gets applied to the document as a whole, or if you are splitting the
document into sections, the first section.

> If the metadata assigned to a document in metadata.xml only applies to the
> whole document, then the functionality of field search of MGPP would be
> significiantly limited, isn't it? For instance, We can't make Greenstone to
> do a Creator and text field search at section level in the desirable way.
> Becasue the Creator Metadata only applies when you do the document level
> search, if this piece of metadata is assigned in metadata.xml. (and if
> that's the case, does it mean we need to put the same creator metadata in
> each section in order to make it work?)

you are quite correct. unfortunately mgpp cant currently handle a search such
as "find sections with snail in the text and Smith in the author of the
document." its much better suited to bibliographic type collections where there
are no sections, and all metadata is at the same level.

to do this, you could either add the metadata to each section, or perhaps edit
the source code to filter the search results. eg
do the section search for the query terms, then check each one to see if the
author was correct before including it in the results list.
the first option would be the easiest but you end up with duplicating a lot of
I guess the best solution would be to change mgpp to handle these types of
queries. maybe one day...

> My another questions is that if I use metadata.xml with Recplug
> -use_metadata_files in configuration, does it mean that I can't insert any
> other section level metadata in the document? I tested, and it doesn't seem
> to work? so I can get around with this dilemma?

I have just done a test, with metadata.xml file, and some Section stuff in the
document (and -description_tags for HTMLPlug). It appears that the metadata.xml
metadata is only used if there is not already some from the document.
eg I specified subject meta in the metadata.xml file and that got added to the
document. I specified Title metadata in both the doc and the metadata.xml file,
and it only used the one from the document.
is this the behaviour that you are having?
You can use the metadata.xml file to add metadata to the document, and then use
the other mechanism to add metadata to the sections.

hope this helps,
Katherine Don