- From: Alan Kent <ajk@mds.rmit.edu.au>
- Date: Mon, 25 Feb 2002 10:29:12 +1100
- To: www-zig@w3.org
Not picking any mail in particular to respond to (wow, where did this thread come from! :-)... I agree with the majority of mails going around. So I thought I would throw in in 2 cents worth. I don't think any single protocol is the right way to go. I think there are two separate problems: the data to be collected, and the technology to make that data available. (I have actually come to this opinion after listening to a Digitial Libraries seminar locally where the real cost turned out to be in collecting the data, not the technology. Then the data had to live beyond a single system as libraries wanted to keep data around for 100's of years. Software does not last that long. So I think its a good prinicple to keep the data and its format separate from the technology using it.) For the data to be collected, I see something like an XML schema describing targets (IP:port, title, description) being a good thing. Maybe even dublin core with an extra element or two (or EAD or whatever). Extending and existing format means less arguing! The only extra information surely over other types of metadata is the Z39.50 host and port. Then there is the issue of how to author and distribute it. I think this will change over time. I don't think Explain is the solution. (I don't think any single technology is the solution actually.) I think a crawler might be able to build up the Z39.50 metadata record by looking into Explain of a server, but that is a separate issue. (No Z39.50 Australian site I have found so far supports Explain by the way.) OAI, UDDI, Z39.50, etc are are possible ways to send the data around. I personally don't think this will be solved quickly. I think people we end up doing what they are able to do (or have funding to do etc). So I think the first step is to define the XML structure for capturing the information. Then people can do lots of technology experiments without having to recollect the data each time. For example, if IndexData put it's list of sites up as an XML document accessible via HTTP, it would be wonderful for me. I can then easily automatically grab a copy, look for the *.au sites, and build the list of Australian sites that I wanted. I probably wont bother with the current HTML files because there is no guarantee that they won't change in format. How to manage and keep the data up to date I think should be a separate thing as, I suspect, it will take much longer to agree on. (I am thinking about union cataloging etc - surely all the issues and politics are quite similar, although the scale is smaller). Alan
Received on Sunday, 24 February 2002 18:29:51 UTC