RE: no Versioning or Variants

I think it is much more likely that variants of a document's content 
would be used for different formats of the same information,
and that all the versions would be in the same language.
For language translations of a document, the most
appropriate thing to do is to make a separate document
for each language, each with its own content variants. For example,
consider the title property. When you translate a document
from English to Norwegian, you translate the title as well.
What, then, is the title of the document? The title of a 
document is usually a "hard index" property (as opposed to
a full text index property), and you search for it with
a traditional "hard property" search, not a full text search.
The obvious fix to this problem is that you simply 
make them be different documents. The English 
version of the document has its title property value in 
English, and the Norewgian version of the document
has its title property value in Norwegian.

As far as content searching of variants, I suspect that
the system administrator indexed one or more (usually
one) of the variants in the full text index, and not the
others. For example, for an image document, he
indexed the OCR text, not the image variant itself.
As another example, he indexed the PDF version of
the document, not the postscript variant or the Microsoft
Word variant.

And when you do a full text query, you get whatever
hits you get on whatever variants of whatever documents
were indexed. The selection of which variants were indexed
does not affect what you get back in the 
result set of your query: You simply
specify the variant or variants you want
in your result set, and you get those variants.

Alan Babich

> -----Original Message-----
> From:	Judith Slein [SMTP:slein@wrc.xerox.com]
> Sent:	April 17, 1998 12:39 PM
> To:	Jim Davis
> Cc:	www-webdav-dasl@w3.org
> Subject:	Re: no Versioning or Variants
> 
> HTTP 1.1 defines enough about variants that I think we need to
> consider
> them.  Section 12 on content negotiation, together with the Accept,
> Accept-Charset, Accept-Encoding, and Accept-Language make it clear
> that a
> single resource may be available in different variants that could
> certainly
> be relevant to search.  Do I want you to search all variants or just
> the
> English language variants or just the French language variants for the
> word
> "chat"?
> 
> At 04:01 PM 4/16/98 PDT, Jim Davis wrote:
> >As I recall, at the IETF BOF we agreed to remove any mention of
> versioning
> >or variants from the DASL spec until such time as the WebDAV protocol
> (or
> >some other relevant protocol) defines them.
> >
> >
> >
> Name:			Judith A. Slein
> E-Mail:		slein@wrc.xerox.com
> Internal Phone:  	8*222-5169
> Fax:			(716) 422-2938
> MailStop:		105-50C
> Web Site:    http://www.nde.wrc.xerox.com/users/Slein/slein.htm

Received on Monday, 4 May 1998 22:47:56 UTC