W3C home > Mailing lists > Public > www-annotation@w3.org > January to June 1999

[Laurent.Denoue@univ-savoie.fr: Re: Yawas : new Web annotation system]

From: Rolf H. Nelson <rnelson@w3.org>
Date: Tue, 23 Mar 1999 18:45:48 -0500
Message-Id: <199903232345.SAA22584@tux.w3.org>
To: www-annotation@w3.org
Laurent gave me permission to forward to the group.  In general, I
personally prefer to discuss these issues on this archived mailing
list, on the grounds that they may be of interest to others besides
myself and Laurent.

-Rolf

------- Start of forwarded message -------
Date: Thu, 18 Mar 1999 23:46:27 +0100
From: Laurent Denoue <Laurent.Denoue@univ-savoie.fr>
X-Accept-Language: en
To: "Rolf H. Nelson" <rnelson@w3.org>
Subject: Re: Yawas : new Web annotation system
Content-Type: multipart/alternative;
 boundary="------------6DBCC0851DFD4DB9948385AA"


- --------------6DBCC0851DFD4DB9948385AA
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

"Rolf H. Nelson" wrote:

> I too would recommend you use a different syntax then your "extended
> url" syntax.  Using something like
>
>    http://localhost:3128/http://www.yahoo.com?note=selected
>
> Has the disadvantage that this "extended url" is not a valid url,
> according to the url standard.  This means there is no guarantee that
> future browsers will be able to handle extended url's.
>
> Therefore at the very least you will want to escape any invalid characters
> using the %xx notation.
>

Yes, I agree. Yawas already escapes special characters like spaces...
Also, I know my extended URLs are not a nive solution.
Requiring every URL to be preceded by http://localhost:3128 is not that good.
What would be good is to use the idea of XPointers, but extending them so as to
include other data, not only the anchor point of the XPointer.

An extended URL should also use # instead of ?.
In Yawas, I've used ? since current browsers don't send to the server the anchor
part of an URL.
But Yawas needs to get this part, so I've used a ?.

In my opinion, an xURL should ressembles XPointers, like :
http://www.yahoo.com#note(word1,word2,wordN)date(19990318)comment(I%20am%ok)...

Also, since anchor points could include a lot of words, I would suggest that
instead of listing every word,
one should only encode the first word, then the number of words, and finally the
last word.
It should be enough to make it unique in one document.

Currently, you can export annotations from Yawas using a "Copy" command : this
creates an xURL in the form I've just
proposed.
So you can paste the xURLs anywhere (email, web pages, word documents,...).
To see the annotation, you first need to "import" this annotation in Yawas.
Yawas will have an "Import" button which will scan the content of the clipboard
for any xURL.
All the xURLs which are found will be added to the annotation database and then
could be visualized on the browser by clicking "View".
This is a little more complicated than directly clicking on the XURL, but since
current browser don't understand XURLs, this is a first solution.

I wish I had time to write an extension to IExplorer so that it can understand
xURLs.
I've written Yawas as part of my PhD thesis where I needed a system to highlight
Web pages. Yawas works well as a research prototype for both Netscape and
IExplorer on Windows (earlier versions worked with everything but did not
support frames).

Here is a screenshot of the current Yawas interface where 3 annotation for page
http://java.sun.com/... have been selected.

[Image]

Here is the result in a Netscape window once the user has clicked on "View" in
Yawas (see above) :

[Image]

I think current bookmark programs have a big problem in that they only let users
keep URLs, not the information
read by users in those pages. I'll try to show that an highlighting tool could
help people retrieve information in their own information space, and better
browse it. It could also help people to cooperate : where today they send URLs,
they could send xURLs and the recipients could see the highlighted parts which
are of interest.

Do you think that extending XPointers with other datas is a reasonable idea ?
In my opinion, important metadata could be :
- -the document type (paper, poem, source code...: see Dublin Core working group)
- - the comment type.

Best regards,
Laurent.
Received on Tuesday, 23 March 1999 18:45:49 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:16:54 UTC