[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Metadata: Sample use cases and classification
I'd also like to offer a "thank you" to Patrick Durusau for chairing the metadata subcommittee. Since the first step is to identify use cases and classifications, here are a few use cases and a trivial metadata classification which I offer to the metadata subcommittee. I hope this email may be useful to the subcommittee as it starts out. First, the MOST IMPORTANT use case to me -- and possibly to many others -- is to greatly improve handling references & citations. I want to have automatically generated bibliographies based on what I happen to cite, in whatever the format du jour is. Current implementations get tantalizingly close, yet don't quite make it. This isn't rocket science; there are existing systems that do this (bibtex, etc.). Yet having this capability would put OpenDocument systems FAR beyond what current typical word processors can do, and would a "killer app"-type reason for many people (particularly in research and academe) to switch. Another use case is to allow far more information about the document itself to be stored inside the document; already we have author, etc. I believe Dublin Core is already supported completely; if not, that'd make sense. ISBN and ISSN numbers would be sensible to store. HOWEVER, I think this should NOT be our initial focus at this time. Twelve years ago I would have said this was the most critical kind of metadata, since finding stuff used to be so hard. But Google and other search engines have become so good at finding things that the need for self-identifying metadata simply isn't as pressing as other needs. If someone wants to work further in this area, well and good, but don't let that impede in any way work on improving automatically-generated bibliographies. Another use case, though far less important I think, would be "security classification" of paragraphs and headings. Some government documents have paragraphs of different classifications (Unclassified, For Official Use Only, Confidential, Secret, Top Secret) with possible categories as well (e.g., REL UK=Releasable to UK, WEIRDNAME=only people with the right to know about WEIRDNAME stuff can see it). Then you can do stuff like "only show unclassified material", etc. Usually each paragraph is marked at the end, e.g., "(U)", and each heading at the beginning, e.g., "(S)". This use is specialized, but it's useful as an example of a different kind of metadata. As far as classifications of metadata go, I can see two major categories of metadata, each subdivided into 2 major subcategories: * Data about THIS document + Data about this document as a whole (author, etc.). This is useful for aiding search. + Data about specific sections of this document (security classifications, etc.). * Data about ANOTHER document + Data about another document as a whole (author, etc.). The bibliography/citation stuff is an example. + Data about a PIECE of another document (e.g., XLink etc. to allow selection or transclusion of a piece of another document). Since the above is a classification of metadata, I could tongue-in-cheek call the above my metametadata, and if different people have different approaches to classifying metadata, we may need to classify the classifications (creating a meta-meta-metadata). Yes, I'm abusing the terminology, but I couldn't resist :-). More seriously, I envision these sorts of proposals to be discussed and worked out inside the metadata subcommittee. Hopefully this email will help things get started. I also think that this shows that the metadata subcommittee charter is a sensible one, since it IS possible to identify use cases and classifications. --- David A. Wheeler
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]