Bijan Parsia's (experimental) Publications Exhibit

The data are all scraped from my DBLP page using a quicky XSLT 2.0 (2.0 for the document function) scraper stylesheet. You have to apply the sheet to a Tidized (i.e., XHTML) version of your DBLP page. It, however, will Tidy up each of your Bibtex pages (via a Tidy online service). If you feed it the original value of your DBLP page via the dblp-page parameter (and you figure out how to pass in the named template, dblp-scrape), it will Tidy that as well. I convert the resulting Bibtex with Babel to the format Exhibit uses (basically, I just followed the "How to set up a publications Exhibit" tutorial).

If Babel's GET interface ever supports (large) embedded strings for the source file, or I can post from XSLT in anything like a standard way, then we could build the whole Exhibit in one step. That would be nice! On the other hand, it would be really easy to bundle up this functionality and provide a publication-exhibit-from-dblp (and maybe other sources) service. But, I'm not doing that!

DBLP gives "nice" Bibtex in that it separates out the publication venue from the particular publication. However, this means that all the "proceedings" the "inproceedings" items refer to show up. Bleah! I'll have to tweak the default parameters not to show them.

Also, it's not, afaict, very accessible. Indeed, there doesn't seem to be anything on the page until the Javascript gets going (you don't get any entries in Lynx, for example). Graceful degradation, folks! (You can get the generated HTML from the copy all button, so that's maybe a good place to start. In fact, that might help the really slow start up time.)

There are some relevant tips for accessibility on the "Making Exhibit Search Engine Friendly" page. From that, <noscript> seems like the best choice for accessibility purpose. Ok, I've added that. The generated HTML is, well, hidious, and the solution overall feels wrong, but at least you can get at the content in Lynx, even if you have a pretty bizarre leftover bit up top. That leftover bit could be generated as well from Javascript, I imagine. It seems to mean that the duplication is, in principle, unnecessary. You could twiddle your default settings, generate the initial HTML, cut and paste that in, and then let the Javascript start from that. Dynamically creating all the items on each pageload is kinda slick, but I'm going to have to do this work around anyway to make the page accessible. Eww, I forsee Javascript programming in my future. How sad.