W3C home > Mailing lists > Public > www-rdf-interest@w3.org > August 2002

Re: RooDolF Update (Google Result in RDF)

From: Seth Russell <seth@robustai.net>
Date: Mon, 12 Aug 2002 13:08:00 -0700
Message-ID: <00a601c2423b$f67a2960$657ba8c0@c1457248a.sttls1.wa.home.com>
To: Alexander Löser <aloeser@cs.tu-berlin.de>
Cc: <www-rdf-interest@w3.org>

From: "Alexander Löser" <aloeser@cs.tu-berlin.de>

> May I ask you some questions:
> *What are you doing with the Google RDF Description in your SailorAgents
> Framework?

Since Sailor Agents only store information in labeled directed graphs,
getting this information in RDF allows me not only to view the information
in a user friendly manner, but also store it if I choose.  But I don't use
your web interface, I just use your URI  which I simply paste into the read
line of my sailor and a user friendly screen of the data appears; but now
the data from the google search is corrolated with all the other data in my
universe.   You should install a sailor agent yourself, its very easy,
especially if you already have Python 2.2 on your system.

> *How could you place messages in my std err screen?

Well I could send you messages via http ... just give me the format and I
will attempt to comply.

> *What's next after RooDolF?

Hmmm ... well here is my wish list:

1) Let me specify which google to use, German or English
2) Give me a way to specify either a keyword search (which you do now) or a
exact match search like you get from the advanced search.
3) Let me specify the number of items to return, I might want 10 or 100 or
5) More compliance with standard terms - in particular dc:title
<http://purl.org/dc/elements/1.1/title> rather than RooDolF:title
6) I would like to be able to click on RelatedInformationPresentURI in my
sailor and actually get related information, either as html or as rdf.
Right now that doesnt seem to work .. not sure whether it is sailor or your
web service.

> In a second step query transformations between popular rdf Query languages
> (Squish, RDQL, RQL, QEL...) and the http interfaces of the rdf wrappes
could be
> established. That's nothing PHD thesis like but it would support the
> with a lot of RDF Data "to play" with. What do you think about this?

Well I am of the opinion that all web services should have an option to
return RDF; then, fortunately (or unfortunately depending on your point of
view) your web service would not be necessarty.  But untill that's the case,
there should be a great demand for converting web services to RDF.  I would
suggest that you make a standard service, so that given the uri of a web
service, and a RDF description of that service and it's inputs, there would
be an algorithm to get RDF out of it with your standard interface.  If you
do it that way, then you don't even need to maintain a web page for each
different service.  For that matter you could just publish the RDF
descriptions of the web services and a python software module and I wouldnt
even need to bother your server.

Seth Russell
Received on Monday, 12 August 2002 16:08:39 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:44:37 UTC