<br><div><blockquote style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;" class="gmail_quote">That working format is currently very close to doing everything that a<br>bibtex-based format can do. I have code in BibDesk that generates that
<br>working format and reads it in and translates it to bibtex.</blockquote><div><div><br>I don't see how either of those statements can be true. For example, BibTeX incorporates a lot of knowledge and constraints about what kinds of documents can be cited and what fields mean for those cases, but Straw doesn't define that. Furthermore, Straw ignores the issue of markup (math, chemistry, mixed scripts/styles, etc.) in citations. And Straw attempts to enforce the use of semantic markup for fields like dates. As a result, different converters might convert BibTeX->Straw and Straw->BibTeX inconsistently, conversions may not render correctly, and, worse yet, converters often probably won't even notice when they're altering citations. In practice (and I've seen this a number of times), converters just end up handling the common cases and leave the rest to manual cleanup.
<br><br>The basic problem with Straw is that, at the same time as defining how to encapsulate citations in HTML, it introduces yet another citation format with yet slightly different semantics from all previous citation formats; no citation manager understands Straw semantics and there is no existing practice. I think encapsulating BibTeX and other formats in microformats is a better approach, since existing citation managers already know how to deal with BibTeX semantics and the problem reduces to one of syntactic and markup conversion (which is hard enough).
<br><br>In any case, in the short term, we've already independently decided to go with Dublin Core for the document metadata for the current collection of documents we're dealing with (and DC already has well-defined embeddings in HTML), so this issue is less pressing. But in the long term, when we will be recognizing and extracting citations from OCR'ed papers, it will be impossible to fully conform with a format like Straw.
<br><br>I'll add some more comments to the Wiki.<br><br>Thomas<br></div></div></div>