W3C home > Mailing lists > Public > public-sws-ig@w3.org > March 2004

Re: A Question about Agent-based Service Discovery/Selection

From: Yuxiao Zhao <yuxzh@ida.liu.se>
Date: Sat, 06 Mar 2004 20:30:13 +0100
Message-ID: <404A26C5.2030207@ida.liu.se>
To: public-sws-ig@w3.org

Your question is (if I understand it correctly):

two kinds of contraints exist for an agent to discover and execute 
automatically a WS: (1) constraints for profile/function/process; (2) 
constraints for communication/access to the WS. Which constraints should 
be checked first? You prefer the latter.

My comments:

1. hard to say which should be first or combined to use, depending on 
quality and speed of filtering, but I tend to prefer the sequnece of (1) 
then (2), because (1) is business requirements and search engines are 
more likely to be available and mature on market and (2) is more 
technical and type validation/matching is often hard and time-consuming.

2. it is important to know that your agent must have a strong capability 
to adjust itself to communicate with those found WSs. Let's assume your 
agent be a WS trying to discover and access to BookAirlineTicket, though 
  any ways (UDDI or search engine) your WS has found two WSs:

WS1: op1(inXSD1, outXSD1), in its WSDL one operations op1 is defined by 
two message types inXSD1 and outXSD1;
WS2: op2(inXSD2, outXSD2).

Your WS in WSDL must define two operations op1(outXSD1, inXSD1) and 
op2(outXSD2, inXSD2) to access to WS1 and WS2, or
your WS must have such an ability to flexibily transform op(inXSD, 
outXSD) into op1(outXSD1, inXSD1) to access to WS1 and op2(outXSD2, 
inXSD2) to access to WS2. If many WSs have been found, the latter would 
be more useful.

Yuxiao Zhao
Received on Saturday, 6 March 2004 14:31:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 9 December 2014 23:03:41 UTC