MultiAgent System Development through SociallyBased Design

1
Agent based Methodology meets Semantic Web
Services
(a preliminary report)
presenter Loris Penserini
penserini_at_itc.it
http//sra.itc.it/people/penserini/
A joint work with
Paolo Bresciani ITC-irst
2
Motivations
The Semantic Web Services approaches are
promising in order to make the service
annotations not only machine readable but also
machine understandable Daml-s
Owl-sLi03Pa02To03.
In this perspective, Ontologies play a key role
in order to describe the application domain.
3
Motivations
Indeed, several techniques have been proposed for
service discovery, composition, and execution.
Nevertheless such techniques have to cope with
the same issue of the domain ontology definition.
Moreover, such approaches do not strongly concern
about how the domain ontology concepts and their
relationships are related to the real
stakeholders needs and responsibilities.
4
Goals
Taking advantage of an agent oriented
methodology, Tropos, we adopt the requirements
analysis and architectural design phases to
characterize the service ontology.
In particular, we give some examples to show how
and when the methodology high-level concepts
drive the modeling of the service ontology
profile in order to develop an advertisement and
discovery service.
5
The Agenda

• - An overview of the application domain
• System requirements analysis
• Architectural design
• Service ontology characterization
• - Services, queries, constraints, and
  advertisements modeling
• Enhancing service discovery
• - Design-time perspective
• - Run-time perspective
• Conclusions and future directions

6
The Application Domain
The application domain deals with such main
stakeholders and their principal dependencies.
7
The Application Domain
The Customer needs drive the Retailer modeling.
8
The System Requirements
The system-to-be has to provide a PC-selling
service, allowing the Customer to increase her
acquaintance about the retailers that can satisfy
her needs.
Some Requirements

• The Customer has to delegate to the System its
  internal softgoal Locate the best retailer
• The System has to know the Customer needs, e.g.,
  what does it means with softgoals PC be correct
  and PC be quikly available
• The System has to know the Retailer offers in
  order to choose the best one.

9
The System Requirements
10
System Requirements towards Architectural Design
The Web-users play a pivot role in the service
providing process, indeed, they ask the system
for services, i.e., by the customer query.
The methodology has to deal with the system
components (new actors) characterization in order
to correctly delegate and perform the identified
system activities Get customer needs, Keep
update retailer info, and Service Searching.
In particular, the proposed architecture is
composed of three system actors User Interface
Manager, Matching Service Manager, and Ontology
Repository Manager.
11
Architectural Design
Goal analysis on the system actor Matching
Service Manager in order to figure out its main
capabilities.
12
The Agenda

• - An overview of the application domain
• System requirements analysis
• Architectural design
• Service ontology characterization
• - Services, queries, constraints, and
  advertisements modeling
• Enhancing service discovery
• - Design-time perspective
• - Run-time perspective
• Conclusions and future directions

13
Dealing with a Semantic Web approach
According to a Semantic Web approach, in order to
make operative the Matching Service Manager
capabilities, we need to define the service
ontology.
Therefore, we aim to adopt the (Tropos)
requirements analysis and the architectural
design results to drive the DAML-S service
profile ontology definition.
Notice that, since this is a preliminary work and
for the sake of simplicity, the ontology examples
are based on a description logic syntax (SHIQ(D)).
14
Driving the Ontology definition
Fragment of the service profile descriptions
Example of customer query
15
Driving the Ontology definition
Example of PC-HW specifications
Example of retailer advertisement
16
Supporting Service Discovery
According to the several proposed approaches,
service discovery is conduct by subsumption
algorithms between the Web-user query (Q) and the
service provider advertisements (A)
Pa02To03Li03. For example, in the Li03
work, the authors propose an algorithm with the
following matching degrees
Therefore, considering our example scenario

• exact, when the query and the advertisement are
  equivalent (Q ? A)
• plugin, when the advertisement allows the
  customer for more service information (Q ? A)
• subsume, when the advertisement is a sub-set of
  the query information set (Q ? A)
• intersection, when the intersection is not an
  empty set (?(Q?A? ?))
• disjoint, when the intersection is an empty set
  (Q ? A ? ?).

17
The Agenda

• - An overview of the application domain
• System requirements analysis
• Architectural design
• Service ontology characterization
• - Services, queries, constraints, and
  advertisements modeling
• Enhancing service discovery
• - Design-time perspective
• - Run-time perspective
• Conclusions and future directions

18
Enhancing Service Discovery
Example 1 (Design-time perspective). Let us
assume that the designer changes the customer
needs throw the related diagram, in particular,
she further details the goal Match H.W.
constraints as illustrated in figure.
As already explained, such a requirement
modification drives the query formalization
hence, let us consider a new query Q2 that fits
the new requirements (still now by hand).
19
Enhancing Service Discovery design-time
perspective
Each time a new concept is going to be added in
the ontology, it needs to be classified in the
terminology hence, a (semantic) consistency
check is automatically performed too.
Notice that, according to our example, Q2 does
not specify an order with a valid PC, i.e.,
NeededPC2 does not refer to a valid assembled PC
(as declared in the ontology by ValidPC1).
20
Enhancing Service Discovery design-time
perspective
Therefore, in such a case the system can give
feedback to the designer (at design-time), e.g.,
suggesting the following modification to the
diagram fragment
21
Enhancing Service Discovery
Notice that, at design time, often the designer
can not model all the possible scenarios that can
arise from the domain, e.g., all the alternatives
in order to achieve a specific business goal.
Example 2 (Run-time perspective). Let us assume
that, for the first time, the system has to
satisfy a new query Q3 that differs from Q1 for
few constraints as detailed in Figure.
New query constraints may reflect, on the
customer needs, a new business strategy
22
Enhancing Service Discovery run-time perspective
In particular, in our example, Q3 cannot be
satisfied over the local knowledge base hence, a
local failure occurs.
That is, new needs may trigger local failures
that the system can overcome by means of its
capability Search for new acquaintances. This
latter architectural requirement can be
implemented taking into account several
approaches, e.g., see Pa01Pe03 and
Li03Pao02To03.
23
Enhancing Service Discovery run-time perspective
Let us assume that, after a collaboration, the
system gets as best match the advertisement
AdvRentOffer1 of Figure that could enrich the
local service profile.
For example, such new ontology elements give
feedback to the customer model in order to
consider also a (softgoal) renting phase to
achieve the goal Obtain a PC.
24
ConclusionsFuture Work
The proposed approach aims to relate the
stakeholders needs with the service profile
ontology of the DAML-S/OWL-S specification.
Moreover, the proposed approach aims to cope with
a need-changing principally in two ways

• at design-time by the user a change in the model
  is propagated down to the ontology
• at run-time by the system a change in the
  ontology is propagated up to the model.

25
Some References
Daml-s The DAML Service Coalition. DAML-S
Semantic Markup for Web Services. Available at
http//www.daml.org/services/. Li03 L. Li and
I. Horrocks. A software framework for matchmaking
based on semantic web tech-nology. In Proc. of
WWW-2003, 2003. Owl-s The OWL Service
Coalition. OWL-S Semantic Markup for Web
Services. Available at http//www.daml.org/service
s/owl-s/1.0/. Pa02 M. Paolucci, T. Kawamura, T.
Payne, and K. Sycara. Semantic matching of web
services capabilities. In Proc. of the 1st Int.
Semantic Web Conference (ISWC-2002), pages
333347. Springer-Verlag. Pa01 M. Panti, L.
Spalazzi, L. Penserini, Cooperation Strategies
for Information Integration, in Proc. of the
Sixth Int. Conference on Cooperative Information
Systems (CoopIS), Springer Verlag, pp 123-134,
Trento, Italy, Sep 2001. Pe02 Penserini, L.,
Lin, L., Mylopoulos, J., Panti, M., Spalazzi, L.
Cooperation Strategies for Agent-Based P2P
Systems. Accepted in WIAS Web Intelligence and
Agent Systems An International Journal,
Publisher IOS Press, ISSN 1570-1263,
2003. To03 R. F. Tomaz and S. Labidi.
Increasing matchmaking semantics in intelligent
commerce sys-tem. In Proc. of IEEE/WIC
International Conference on Web Intelligence
(WI03). IEEE Press, 2003.