[uf-discuss] hCard to represent simple entities

Jim O'Donnell jim at eatyourgreens.org.uk
Sat Jan 5 08:54:34 PST 2008

On 5 Jan 2008, at 16:07, Andy Mabbett wrote:

>> Sorry, I should have been clearer. What I want to do, in terms of  
>> marking up content, is determined by how people are going to use  
>> the web site.
> Yes, but you don't know how people are going to use it.
Yes I do. If I'm building an archive of documents related to maritime  
history, I've got a pretty good idea of how historians and amateur  
geneologists are going to use it. I don't know what sort of  
applications people might develop on top of our data, and I think  
that's where opening up museum collections gets really interesting.

>> If people want more intelligent searches - 'show me  manuscripts  
>> written by Captain Cook' - then rel-tag seems like the  natural  
>> tool for marking up names.
> Suppose I want to know what Captain Cook drank; and that your  
> manuscript includes a letter from Cook to a friend back home,  
> saying "I don't miss much of home, but I could murder a pint of Mrs  
> Hecklethump's famous beer"
...and the text referencing Cook within the letter, or article, is  
"James Cook" or "the captain of the Endeavour" or simply "James" but  
not "Captain Cook".

> So I need to be able to specify that I only want pages which refer  
> to a person called "Captain Cook" - in other words, who has an  
> hCard whose "honorary-prefix" is "Captain" and whose "given-name"  
> is "Cook". [One day, I might also be able to search for "hBeverage"  
> or some such (perhaps a subset of "hRecipe") instead of "beer", too.]
Or, better, you want a search which finds all text tagged as  
referring to a person whose canonical name is "Captain James Cook."  
That way, you don't have to worry that your search term needs to  
exactly match the text, only the tag. In fact, a clever enough search  
could probably find all tags which are names of people and are close  
matches to your search term, then find all the documents tagged with  
those tags. Something like this search I put together for our prints  
Tagging can handle this if there's a mechanism in place to put some  
context on the tag - indicate that it refers to a person, a place, a  
business or whatever.

> That's how using hCard adds re-usable semantics to a person's name,  
> even if that's the only "vcard" data you publish about them.
Yes, but aren't you assuming that Cook's name, including rank, will  
be present in the text, which won't be the case. Cook didn't sign his  
letters "Captain Cook". He may well have signed personal letters  
simply as "James".

>> I don't  see a use case for getting the contact details of Captain  
>> Cook.
>> hCard does a specific job very, very well - it enhances social  
>> networking. I'm struggling to see, though, how it generalises to  
>> marking up the names of all people, living and dead.
> As explained previously, hCard is not /just/ for contact details;  
> it's "for representing people, companies, organizations, and  
> places". That's quoted directly from its spec. And that's how it's  
> already being used, "in the wild".
Are there any examples where it's being used for something other than  
social networking?

>>> For instance, adding a tag doesn't tell a future search engine  
>>> that your text is about a person.
>> I think the answer to this is the rel-tag microformat coupled with  
>> sensible URLs to give much more intelligent tags. Imagine if the  
>> Maritime Museum archives used tags like:
>> <a rel="tag" href="/search/person/Emma_Hamilton">William  
>> Hamilton's wife</a>
>> <a rel="tag" href="/search/vessel/HMS_Victory">the Victory</a>
>> <a rel="tag" href="/search/person/Captain_James_Cook">Captain  
>> Cook</a>
>> There's enough info in the HTML there to allow for some quite  
>> intelligent searching.
> Only if you can get everyone to use the same URL structure, and to  
> make all the subjects in their content into links.
Yes, that's why I'm interested in a lightweight, flexible solution  
that could be applied to archives in general, not just the holdings  
of the National Maritime Museum.

>> Classes distinguishing between the different  types of tag could  
>> be added to the links too.
> Indeed - and the appropriate class, in the latter case, would be  
> "vcard", with "fn, "honorary-prefix" and "given name" around the  
> content as per the hCard spec. Why would you want to invent a new  
> set of classes?

I don't, if hcard will do the job. But I also have to be able to mark  
up the phrase 'William Hamilton's wife' with the name 'Emma  
Hamilton', using the same set of classes, to indicate that she's also  
a person. hcard applies to one very specific subset of reference  
strings, those which are the exact names of people, but not to  
reference strings in general. A solution that doesn't apply to all  
references to people may not be very useful to this particular problem.


Jim O'Donnell
jim at eatyourgreens.org.uk

More information about the microformats-discuss mailing list