General developer forum

Here's another thought on using the XML metadata from IMS/SCORM
and a full text search engine to create a Google/Yahoo like web directory.

The IMS Content Package is a file directory tree + imsmanifest.xml.
For exchange between systems, this is zipped up.
Within a digital respository (edukalibre), just use in the unzipped form.
That would fit nicely with SVN (or CVS).

Within the imsmanifest.xml, there is a structure that maps to the
files as resources. One can also associate metadata here, within this
XML. This is the IMS Metadata, which is a superset of DC.

Next, run the Lucene full text indexer over the imsmanifest.xml.
You can get Lucene from Apache. This is a commercial grade full
text search engine, the originator took many of the ideas from Excite
(which for a time had an open source release).

You can run the Lucene indexer DIRECTLY on the imsmanifest.xml.
Now you've got a full text database of all the metadata. I've done this before
with many other full text indexers, like Verity and Infoseek/Ultraseek/Inktomi.
Or freeWAIS. While one typically thinks about doing this on HTML, it will
work just fine on any old XML.

From here, you'd have to do some work to figure out how to formulate queries
and creating browsable directory listings. This is somewhat like Excite, Yahoo,
or Google directories. The Lucene query engine has both full text and boolean search capabilities.

In a local project here at UW-Madison, we are about to use Lucene to full-text
index an image collection of Art History (doesn't everyone have such a collection :-0) http://www.wisc.edu/arth/insitu/index.html
The current search is plain old SQL.

Looking as some of the example Lucene sites, I see lots of very interesting examples. Done right, this would do everything that Harvest Hive is doing B->