Nice post, on a classic issue for information systems. Just a note, I’m not sure @sshreeves would want to be associated with our logic-chopping extremism. But I’m quite flattered to be confused with her at any rate.

I agree entirely that we need both approaches if we want the semantic web to do the things we hope it will do. We were trying to draw attention to the unintuitive consequences of a simplistic approach based on mathematical notions. But developing a fully correct ontology has it’s own challenges.

-Karen.

Part of the appeal, for me, of Levy’s approach is that he really isn’t trying to express a theory of documents, which can then be acted upon using logic (presumably by a computer) … His goal instead is just to explore just how rich and shifting the Document landscape is, to get a sense of the intuitive semantics of “Document”. I guess it’s almost a quantitative/qualitative thing. We really do need both, and it was interesting that @allenrenear and @sshreeves @karenwickett paper ends with:

One has to wonder whether all this logic-chopping subtlety is going to be worth it. It isn’t clear exactly how to finish the job, and yet it is clear that some of our most familiar — and effective — ways of conceptualizing our domain will revised if we continue along this path. Particularly troublesome is the prospect that the revisions anticipated will add not only complexity in design and use, but increase computational complexity as well.

I’d also recommend taking a look at the work of my colleague Allen Renear who has been struggling with these questions for a long time, along with other scholars in the digital humanities. A list of his publications is here: http://people.lis.illinois.edu/~renear/renearcv.html

I’d recommend “What Is Text, Really?” (http://doi.acm.org/10.1145/264842.264843) and “Three of the Four FRBR Group 1 Entity Types are Roles not Type” (http://is.gd/3nC0T) – which is really about how XML “documents” fit into FRBR.

Relevant to this post, I would also complement the bibliography with a long term work made a by a group of scientists in France under the surname Roger T. Pedauque. In particular, http://archivesic.ccsd.cnrs.fr/docs/00/06/22/28/PDF/sic_00000594.pdf is a MUST read.
Cheers.

Raphaël