<div>Norman, You mentioned a bot (a web crawler) which probably needs a gatherer component to 'harvest' metadata (aka 'deep web' crawler) using one of these protocols (CSW, WFS or OAI-PMH). </div>
<div> </div>
<div>What I am currently thinking about here - inspired by OAI-PMH and Dublin Core - is a) do we need persistent metadata identifiers, b) is an attribute indicating protocol enough as an entry point to 'discover and bind' even specific services like WMS and later on SOAP (according to "occam's razor prinziple" or to the parsimonious approoach)?
</div>
<div> </div>
<div>Or did you mean to spidering also through HTML pages as a 'focused crawler'? If yes, I know Heritrix and I am sure there are some tools around at OAI-PMH tools page. What do you plan to use?</div>
<div> </div>
<div>-- Stefan</div>