I won't be there, but a couple of thoughts:
* Address the fact that PDFs are not an open format, and the contents (tables and formatting) can't easily be extracted without expensive software. If an organization wants to publish their work as PDF, but wants to be open, they need to publish the work in a different format as well. Far better, don't publish as PDF at all - just put it on your website.
* Rich text: While databases and text are important, there are times when we deal in rich text, whether on the web, in PDFs or in Word or Open Document Format files. I would suggest HTML for any rich text. (This insight comes after struggling with wordprocessors and the bloated files they produce.) It's easy for anyone to read on pretty much any system. There are a number of good editors that not only create rich text as clean HTML in a WYSIWYG display, but also let you get behind it to the actual code. "If you try to hide the complexity of the system, you'll end up with a more complex system." (Arch Linux design philosophy.)
* Michael Maranda has talked/written about open standards in knowledge sharing (though I can't find it right now - I'll post when I do). His project is Catalytic Communities, for sharing info about development projects, and they are concerned to enable sharing between their project and others such as Appropedia.
Just wanted to chime in that I am no longer part of the org mentioned - though I am continuing the basic work on open standards with special focus on the social benefit sector or better world building... I plan on posting more (at an as yet undetermined location) on the subject in coming months,
I just wanted to add that often when people think of open standards they think of format and not also of content. Addressing key challenges like unit of analysis, currency conversions, place names, etc are key for combining data from different sources. Mostly I've had to do these things by hand because either no standard exists or multiple ones do (e.g. country codes). If there is an easy technological solution to data mashups where different standards exist for content I know quite a few people who would love to see it.