Re: Web page rating service through proxy servers

From: Ian Cooper (ian@mirror-image.com)
Date: Mon Jul 26 1999 - 01:33:41 MDT


On Sun, 25 Jul 1999 20:47:55 +0200, Jacob Palme wrote:

>In the EU-financed research project SELECT (see
> http://cmc.dsv.su.se/select) for more info), we are looking at ways
>of providing rating services on web pages. By rating services we do
>not mean the PICS kind of services, established to protect children
>from unsuitable web content, but collaborative filtering, where
>people store their ratings of web pages and these ratings can help
>other people find the most interesting pages on the web.

There's already been a bit of research into this field (my own PhD
thesis included[ShamelessPlug]). A couple of papers that come to mind
are:

@InProceedings{WebTagger,
  author = {Richard M. Keller and Shawn R. Wolfe and James
                  R. Chen and Joshua L. Rabinowitz and Nathalie
Mathe},
  title = {{A Bookmarking Service for Organizing and Sharing
{URLs}}},
  booktitle = {Proceedings of the Sixth World Wide Web Conference},
  editor = {Michael R. Genesereth and Anna Patterson},
  year = 1997,
  organization = {World Wide Web Consortium},
  publisher = {Elsevier, Holland},
  month = apr,
  pages = {13--24},
  note = {ISBN 0-9657614-0-1},
  URL = {http://proceedings.www6conf.org/HyperNews/get/PAPER189.html}
}

@Article{Coollist,
  author = "Jong-Gyun Lim",
  title = "{Using Coollists to Index {HTML} Documents in the
{Web}}",
  journal = "Computer Networks and {ISDN} Systems",
  year = "1995",
  volume = "28",
  number = "1--2",
  pages = "147--154",
}

@Article{CacheCatalogue,
  author = "Chris Dodge and Beate Marx and Hans Pfeiffenberger" ,
  title = "{{Web} cataloguing through cache exploitation and
                  steps toward consistency maintenace}",
  journal = "Computer Networks and ISDN Systems",
  year = "1995",
  volume = "27",
  number = "6",
  pages = "1003--1008",
}

>One major problem is that we want people to be able to give ratings
>on any web page, anywhere. This can be done either by adding a rating
>button to their web browsers through a plugin, or by letting a proxy
>server add the rating button in a small frame at the top of the web
>page. The advantage with the proxy server solution is that it will
>work for all web browsers. Modifying web browsers with a plug-in will
>only work with particular web browsers.
>
>One problem with adding a rating button in a small frame, is that the
>URL in the address field will be the URL of the frame and not the URL
>of the web page below the small frame. I believe, however, that a
>proxy server can fix this, too.
>
>Of course a proxy server cannot start adding rating buttons to all
>web pages, people would scream at them. The proxy server will have to
>have two addresses, one address for those who do not want to make
>ratings, and one alternative address used only for those who want to
>provide ratings.

I don't think it's quite as simple as that. I'm sure you've already
consisdered this, but one of the biggest problems is in identifying
which content to attempt to attach your buttons to. And then, with
already-framed content, working out how your rating buttons will
relate to the sub-frames (if at all). Further, is any rating shared
among all components from which the page is constructed (e.g.
including the graphics - which are sometimes more important than their
container), or just the container?

>Is there any proxy software developer, or proxy server, which is
>interested in co-operating with our project in adding such a facility?

I can't offer help in that way, but I can offer a few more references
to folks that have developed extensible proxy servers with aims like
this in mind:

@Article{Strand,
  author = "Charles Brooks and Murray S. Mazer and Scott Meeks and Jim
Miller",
  title = "{Application-Specific Proxy Servers as {HTTP} Stream
Transducers}",
  journal = "World Wide Web Journal",
  year = 1995,
  volume = "1",
  number = "1",
  pages = "539--548"
}

@Misc{V6,
  author = "Bernard Lang and Fran\c{c}ois Rouaix",
  title = "{The {V6} Web Engine}",
  URL = "URL:http://pauillac.inria.fr/\%7Erouaix/V6/",
}

I used a (heavily) modified version of the code in sfgate, but that
was mostly because it was in Perl (and therefore hideously expensive
and not particularly useful for anything but some prototype research):

@Article{SFgaate,
  author = "Ulrich Pfeifer and Norbert Fuhr and Tung Huynh",
  title = "{Searching structured documents with the enhanced
                  retrieval functionality of {freeWAIS-sf} and
{SFgate}}",
  journal = "Computer Networks and {ISDN} Systems",
  year = "1995",
  volume = "27",
  number = "6",
  pages = "1027--1036",
}

Another slightly related work is:

@Misc{SIMON,
  author = "Mark J. Johnson and E. H. Mamdani",
  title = "{Feedback In Internet Resource Discovery Systems:
{SIMON}}",
  URL =
"URL:http://web.elec.qmw.ac.uk/simon/irdpaper/ird-paper\_1.html",
  year = 1994,
  month = nov,
}

>(Please excuse that this message may not be appropriate for the wrec
>mailing list. And please, if you write replies to this message, which
>are not related to the issues of the wrec working group, please send
>these replies personally to me, and not to the list. I am not sure,
>but there may be issues related to the problem presented in this
>message which are relevant to the issues of the working group, too.
>Replies on such issues can of course be sent to the list.)

I'm not certain whether this is or is not an area in which WREC will
become interested in any way (with apologies if this is just noise).
Proxies certainly sit in an interesting point within the network, and
therefore see a lot of information which is very useful. It's
somewhat possible that they offer the ability to help index more of
the Web than may be possible by the big search engines' robots - but I
think that's another topic...

Ian

[ShamelessPlug]: @PhdThesis{IHCThesis,
  author = {Ian Henry Cooper},
  title = {Web Resource Re-Discovery: Personal Resource
                  Storage and Retrieval on the WorldWideWeb},
  school = {The University of Kent at Canterbury},
  year = {1998},
  OPTkey = {},
  OPTtype = {},
  OPTaddress = {},
  OPTmonth = {},
  OPTnote = {},
  OPTannote = {}
}



This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:26 MST