[open-bibliography] Basic RDF/XML to bibtex/RIS conversion?

[Jim Pitman]

Peter Murray-Rust <pm286 at cam.ac.uk> wrote:

> > It is always a problem going from an extensible semantic framework to one
> with a fixed vocabulary and no namespaces. There is likely to be semantic
> and content loss. For example the "RIS-like" fields in NIH-RIS do not all
> map onto the fields in RIS (or if they do - e.g. "note" or "id" they lose
> semantics.
> People often try to "mend" this by adding more fields and clever syntax.
> This is doomed to failure - it simply breaks software written to the main
> standard.

For the long term, agreed. However, short term fixes can be useful.  We still do not yet
have adequate tools for editing BibJSON, and there are such tools for other formats. I used 
extended BibTeX for many years as a proxy for what has become BibJSON. It is possible and
may be useful for communication to roundtrip even quite complex BibJSON to BibTeX or
other key-value formats. 
I am dealing with a developer who understands BibTeX and doesnt want to deal with BibJSON.
So fine, I can provide a crosswalk. But as you say, any such crosswalk is likely to break
with someone else's BibTeX extension.

> And that's a primary reason why we developed BibJSON. It can hold anything
> we want. In summary:
> Legacy => BibJSON (no semantic loss, though possibly a need to write more
> vocabulary)
> BibJSON => Legacy (non-zero probability of semantic loss and content loss)

Exactly.

--Jim