W3C home > Mailing lists > Public > www-ws-arch@w3.org > January 2004

RE: Service Discovery comments

From: Katia Sycara <katia@cs.cmu.edu>
Date: Wed, 7 Jan 2004 19:16:27 -0500
To: 'David Booth' <dbooth@w3.org>, www-ws-arch@w3.org
Cc: katia@cs.cmu.edu
Message-ID: <005301c3d57c$ab127380$d1bd0280@scs.ad.cs.cmu.edu>



Here are my comments on the service discovery related parts of the document.
In " " I enclose text from the docment and I indicate desired changes by
enclosing them in *...*. I justify the suggested changes in subsequesnt
comments, where appropriate.
  --Katia
----------------------------------------------------------
1. section 2.3.3.2.1.
 "Discovery is the act of locating *a Web service through its *
machine-processable  description that may have been previously unknown and
that meets certain functional criteria"

Katia comment: the current definition states that the discovery is the acto
of locating a machine-processable description of a Web service ...
The goal is to discover the service rather than its description so I think
that stating this explicitly in the definition is preferable

2. section 2.3.3.2.2 Relationships to other elements

 "Discovery is *realized by*
   Matching a set of functional and other criteria with a set of service
descriptions"

Katia comment: since we do not have a clear definition of an act but we do
use *realized by* quite often, this seems more appropriate

3. section 3.1 Step 2 bullet 3

I propose to get rid of the parenthesis with "(excepting the network address
of the particular service)"

Katia comment: it is not clear why this parenthetical text is there. If we
keep it in bullet 3 we need to add it in bullet 4 where it is currently
missing

4. section 3.1.
The paragraph after Step 4 starting with "The overall process of engaging a
Web service was outlined in the introduction and included the following
steps ..." is redundant and can be omitted

5. section 3.1.1
In 1.b. 
"The requester entity (either a human or a requester agent) specifies
criteria *and sends them to the discovery service to enable selection of a
Web service description* based on its associated functional description and
potentially other characteristics". Etc...

Katia comment: the text in * * must be added since step (c) talks about the
discovery service returns one or more Web service descriptions that meet the
criteria. The discovery service must have obtained the requester's criteria.

6. section 3.1
Step 2 typo (p. 76) 
"Step 2 also requires that the parties agree on the service description that
is to be used. However, since the requester entity obtained the Web service
description in step *1.c* [instead of the current 1.3] in effect ..."

7. section 3.1.3

I do not completely agree that "people are skeptical in allowing machines to
make judgement decisions for them ...." (counterexamples abound, but let us
not get into that discussion). We can augment the current text by saying
that 
"In automated discovery, there are two cases for mitigating the trust issue:
1 agents could autonomously discover services and then, show them to the
human user to choose.
2. Agents autonomously discover services and then the requester agent, upon
receiving the set of discovered services can perform some sort of checking,
for example searching the Dunn and Bradstreet registry for the service
providers' quality rating."

8. section 3.1.4. 
"At present there are *three* leading viewpoints on how a discovery service
should be conceived: as a registry or as an index or as a *peer to peer
process*." 

Katia comment: peer to peer discovery would be useful in ad hoc and dynamic
networks, especially for military applications. We may want to mention the
p2p case here for completeness. If people agree, I can write the explanatory
text.

9. section 3.1.4.

I disagree that UDDI is an example of the registry approach, as defined in
this section ie "A registry is an authoritative, centrally controlled store
of information". Where the next three bullets go on to define what
authoritative and centrally controlled mean. This is because:

a. UDDI registry can be either "external/public" like IBM, Microsoft or HP
UDDI registry or it can be "internal or privately maintained". Hence the
control over registry can be modified by the owner and hence the control of
the publishing behavior can vary from one UDDI registry to another.

b. The only requirement to publish in the UDDI registry is a registration
process (not any different from registering for e-mail service). In the
terms and conditions page, one of the requirement is  for a registrant to
provide accurate information. But after the registration process there is
nothing that stops the registrant from publishing any information it wishes.


10. 3.1.4 (p. 78)  "indices" instead of "indexes" 
and also in the fifth bullet Different indices could *provide* (instead of
*provided*) .....

-------------------------------------------------------------------------
Received on Wednesday, 7 January 2004 19:18:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:24 GMT