Re: DAML-S and the UDDI initiative

I am sorry with all the peple that raised questions about our work
with DAML-S and UDDI.  Unfortunately I was delayed in my return from a
trip, so I just found all the messages.

CHIN, ALVIN writes:

> I've read the paper and have some questions regarding about the
> publishing and requests for services in the DAML-S/UDDI matchmaker.
> If we are doing semantic searches for services using the matching
> algorithm proposed in the paper, does this mean that UDDI does not
> come into the picture after service publishing? (ie. After the
> service is registered with UDDI and then advertised in the
> AdvertisementsDB of the DAML-S matching engine).  According to the
> paper, it says that requests are sent from the Communication Module
> to the DAML-S Matchmaker engine.  This means then we don't do a
> discovery in the UDDI registry, right?  We only do a discovery in
> the UDDI registry if we want to match by keyword in the TModel of
> the DAML-S profile (added to the UDDI record), that is, not do a
> semantic search.
 
I understand your objection, and indeed I remember wondering about it
when I wrote the paper. I am sorry I have not been clear enough.  The
architecture described in the paper represents the system as it was
implemented.  We kept the semantic matching engine module outside UDDI
because we did not have direct access to the source of a UDDI server,
rather at that time we used the public IBM repository.  Ideally, we
could have tried to fetch advertisements straight from UDDI at request
time, while this is a possibility it turned out to be way too slow for
our experiments.  Since then, we had been considering using other UDDI
servers that are open source, but we did not get around to do it yet.
Still, I do not think that this invalidates our claims, because a
"DAML-S enabled UDDI" should be extended to the whole picture, not
just the UDDI module.  UDDI API would also be affected by adding
capability matching, because we should define messages that allow UDDI
clients to request the capability matching and those messages are
outside the messages that UDDI API allows right now.

> I saw your posting from the www-ws@w3.org web site regarding about
> DAML-S and UDDI.  You mentioned that your group was going to release
> a DAML-S for the HP Jena by the end of the month.  I was wondering
> if it would be available for public academic release.  I wish to
> describe web services semantically for discovery and invocation by
> mobile clients, and having access to the DAML-S for Jena API would
> be fantastic.

Unfortunately, because of the delay of my return I was not able to
follow up with the public release by the end of June.  I am a bit
reluctant to release the initial version of DAML-S API right now
because we are planning to release DAML-S 0.7 soon, and it may be a
bit different than the current version (0.6) to accomodate comments
and suggestions that we got from people that uses DAML-S.  Also, I
noticed that a new version of Jena has been released, I had no time to
verify what changes did they make. I still want to release the API,
but it will have to be after we decide on DAML-S 0.7.  In the
meanwhile, if anyone is interested to use the API for DAML-S 0.6
please e-mail me and I will give you what I got at that time.

As for the conditions of the release, I will have to adheere to what
CMU tells me to do.  Basically, it is a standard licence that says
that the software will be free for use but CMU retains the
intellectual property and that there is no warranty and we are not
liable for anything.

I hope this clarifies most of the questions,

--- Massimo

Received on Tuesday, 9 July 2002 19:18:43 UTC