- From: Nick Kew <nick@webthing.com>
- Date: Sun, 31 Dec 2000 18:32:59 +0000 (GMT)
- To: Charles McCathieNevile <charles@w3.org>
- cc: WAI ER group <w3c-wai-er-ig@w3.org>, Dan Brickley <danbri@w3.org>
On Sun, 31 Dec 2000, Charles McCathieNevile wrote: > OK, so I thought for a bit. I'm missing the context for this, but I'll post anyway. > One possible approach is to define a dictionary, somewhere on the Web. Then > you provide a reference to it as part of the metadata for a page. Then a > dictionary lookup system can go dereference things that are not real words > (or things that are - why not...). This is interesting. When I introduced a (primitive) spellcheck option in the Site Valet, it was suggested to me that I should allow users to link their document to a dictionary, which would then be used in the spellcheck. I have hitherto rejected the suggestion, on the grounds that dictionaries are rather big, and I don't like the bandwidth implications of fetching them. One mechanism I might introduce is to allow users to add words to (private per-user) extensions to a system dictionary. Now, regarding your suggestion, I don't see how a standard dictionary somewhere on the web is any better than a standard dictionary on my server. However, a more interesting idea is a Dictionary Protocol, designed to allow dictionary servers to exchange information on-demand. However, I suspect this isn't really the right forum to discuss a dictionary protocol. > Questions: > How do you know something is a real word - from markup or from spelling? > > How do you avoid getting a wrong answer in ambiguous cases, or looking up > something that isn't really an acronym, or not looking up something that is? Now that you build into the protocol: it should include well-defined meta-information that identifies a dictionary response as a definition, an abbr-expansion, a translation, etc. The request would also indicate "I'm only interested in responses of class ????" (c.f. HTTP Accept). > I think the answer to this one is that markup is a lot better. (You could > rely on people using upper case letters, but it doesn't always work. And is > not even correct, as I understand it, in Italian.) E alcuni anni fa quando sono stato in Italia, ma credo che Lei e sbagliato! > This means that a user > agent or add-on could expect to find an expansion of abbreviations that were > marked as such but not explicitly expanded in > the same document, or if not there in the dictionary for the document, or in > the dictionary referenced by the markup for the particular abbreviation. Indeed, everyone could benefit from a dictionary protocol supported by user agents, servers and authoring tools. -- Nick Kew I procrastinate. Efficient people never get around to it. Therefore I am more efficient than an efficient person.
Received on Sunday, 31 December 2000 13:34:06 UTC