Deciding Semantic Matching of Stateless Services

1
Deciding Semantic Matching of Stateless
Services Duncan Hull, Evgeny Zolin, Andrey
Bovykin, Ian Horrocks, Ulrike Sattler and
Robert Stevens
School of Computer Science, University of
Manchester, UK. first.last_at_cs.man.ac.ukDepartmen
t of Computer Science, University of Liverpool,
UK. andrey_at_csc.liv.ac.uk_at_ American Association
of Artificial Intelligence (AAAI) 2006, Boston,
MA, USA
ABSTRACT A novel technique for semantically
matching service requests with advertisements is
described. This technique extends current
approaches by explicitly stating the relationship
between the input and output of a given service.
The meaning of the terms used in the description
is defined using OWL (the Web Ontology Language)
and this has several advantages outlined below.
INTRODUCTION Understanding and managing the data
generated from Human Genome Project is
recognised as a grand challenge for both computer
science and biomedicine. All of the raw data, and
many of the tools to interpret and analyse it,
are publicly available as Web Services. As of
2006, around 3000 highly heterogeneous and
autonomous services are available from within
client applications like the Taverna workbench
1, part of the myGrid project. Current
techniques for matching service requests with
advertisements for biomedical services have
proved inadequate. This poster, and accompanying
paper 2, outlines a novel technique for
semantically matching service requests with
service advertisements. We demonstrate that this
matching technique improves both precision and
recall of web service matching.
EXAMPLES The illustrations below show a sample
advertisement with two related service requests,
using a commonly used standard example of Wine
buying from 3
Service advert INPUT g GeoRegionOUTPUT w
WineTHERE IS SOME f WineGrower f,
LocatedIn(f,g), Produces(f,w)

• RELATED WORK Several related research projects
  use logic-based knowledge representation to
  advertise and find Web Services in registries
• The OWL-S profile desribes IOPEs (Iinputs,
  outputs, preconditions and effects) but
  currently has no way of relating inputs to
  outputs due to limitations in the OWL language.
  Since many of our services are stateless, we do
  not need stateful descriptions that OWL-S
  provides.
• BioMOBY and myGrid advertise services with a
  related language (RDF and RDFS) that is less
  expressive than OWL and cant relate inputs to
  outputs at the class level
• The Web Services Modelling Ontology (WSMO) has a
  mechanism for relating intputs to outputs, but as
  far as we know, no decidability results exist for
  this approach

No match
(graphically)
Service request 1 INPUT g GeoRegionOUTPUT w
WineTHERE IS SOME s Shop s, LocatedIn(s,g),
Sells(s,w)

• The description of the relationship between input
  and output, allows two similar advertisements
  that have the same inputs and outputs, but
  perform different functions to be distinguished
  from each other - this enables more precise
  matching, minimising false negatives.

Match
Service request 2 INPUT g FrenchGeoRegionOUTPUT
w FrenchWineTHERE IS SOME f WineGrower f,
LocatedIn(f,g), Produces(f,w)
The approach described here complements and
extends each of the above. In addition the
service matching problem is reducible to
subsumption of conjunctive queries, hence it is
decidable for many Description Logics 2.
OUTPUT Wine
INPUT GeoRegion
CONCLUSIONS We are currently implementing the
approach described here in a public registry of
biomedical services. Future work will investigate
using this technique to describe and retrieve
compositions of services e.g. Workflows, which
are commonly used by scientists conducting
experiments on genomic and proteomic data on the
web.
The ability to use a background ontology during
service matching, allows us to increase the
recall of matching algorithms by reasoning that
Wines producedIn FrenchGeoRegion are FrenchWines

• REFERENCES
• Duncan Hull, Katy Wolstencroft, Robert Stevens,
  Carole Goble, Matthew Pocock, Peter Li and Tom
  Oinn (2006 ) Taverna A tool for building and
  running workflows of services. Nucleic Acids
  Research 34 W729-W732 (Web Server Issue)
• Duncan Hull, Evgeny Zolin, Andrey Bovykin, Ian
  Horrocks, Ulrike Sattler and Robert Stevens
  (2006) Deciding Semantic Matching of Stateless
  Services. Proceedings of the Twenty-First
  National Conference on Artificial Intelligence
  (AAAI-06), Boston, MA, USA, July 16-20
• Dean Martin et al (2004) Bringing Semantics to
  Web Services The OWL-S approach in Proceedings
  of First International Workshop on Semantic Web
  Services and Web Process Composition (SWSWPC 2004)

Acknowledgements This work was supported by the
myGrid UK e-Science programme EPSRC GR/R67743/01
and GR/S63168/01 http//dynamo.man.ac.uk Dynamo
Project http//taverna.sourceforge.net Taverna
project http//www.mygrid.org.uk myGrid
project .