WEB SERVICE RETRIEVAL BASED ON ENVIRONMENT
ONTOLOGY
Budan Wu and Zhi Jin
Academy of Mathematics and System Science, Chinese Academy of Sciences, Beijing, China
Keywords:
Web Service Description, Web Service Retrieval, Environment Ontology, Semantics.
Abstract:
Existed descriptions of Web Service are in syntax level, and still relies on people to understand and decide
what a Web Service can do, which encumbers automatic and dynamic Web Service retrieval. In this paper,
we propose environment ontology as the base of describing semantic of Web Services, whose resource and
state transitions reflect Web Service’s behavior. We construct the description format based on environment
ontology and put forward a road map approach towards automatic and dynamic Web Service retrieval.
1 INTRODUCTION
The booming of Web Services requires Web Ser-
vices encoded in a description form understandable
to computer. With this goal, the industry put forward
WSDL(E.Christensen et al., 2001) and BPEL(Fu
et al., 2004) which are standards for Web Services
definition and orchestration. However, these stan-
dards are on syntax level using ports and interface
to invoke Web Service functions. A first step to-
ward this problem is to develop formal languages
and inference mechanisms for representing and rea-
soning with core Web Service concepts.(Paolucci and
Sycara, 2003) Thus, in parallel with those efforts in-
troduced above, the Semantic Web(Berners-Lee et al.,
2001) community has been developing languages and
computing machinery. Efforts include RDF(Gutierrez
et al., 2005), OWL-S(de Vergara et al., 2005) and
IRS-II(Motta et al., 2003). Unfortunately, these ef-
forts show limitations in automatic Web Service re-
trieval in practice. Motivated by accurately depicting
Web Services in computer-understandable form, and
specifying constraint on action sequence, state tran-
sitions, we use environment ontology with richer se-
mantics to depict the specific capability of Web Ser-
vices and put constraints on interaction sequences,
states transitions and other related. The classic for-
mula(M.Jackson, 1997) in requirement engineering
expresses the importance of environment(M.Jackson,
2001). Deriving from it, environment changes re-
flect Web Service behavior. When both Web Service
capability and requirement are mapped to the same
environment ontology, a matchmaking can be devel-
oped between them automatically. The originality of
this approach compared to other methods is that we
use environment ontology to describe Web Services
in formal semantic level and put forward a procedure
of automatic Web Service retrieval based on environ-
ment ontology.
The paper divides into 5 sections. After the intro-
duction, environment ontology and an online selling
domain ontology are explained in section 2. Based
on environment ontology, section 3 shows Web Ser-
vice description method. In succession, Web Service
retrieval is expounded in section 4. Finally, a conclu-
sion is made.
2 ENVIRONMENT ONTOLOGY
2.1 Top-Level Environment Ontology
Top-level ontologies provide a set of basic rela-
tions and the classes to express the constraints
on their use. It is represented by: TO :=<
TC, TR, H
TC
, Rel, A
TO
>
• TC and H
TC
are concepts explained in Table 1.
382
Wu B. and Jin Z. (2007).
WEB SERVICE RETRIEVAL BASED ON ENVIRONMENT ONTOLOGY.
In Proceedings of the Third International Conference on Web Information Systems and Technologies - Internet Technology, pages 382-385
DOI: 10.5220/0001275403820385
Copyright
c
SciTePress
The first section of the table represents the re-
source ontology of the Web Service, the second
represents the interaction ontology between Web
Services and environment entities.
Table 1: Some Meanings of the Representation Concepts.
Environment Entity entities which a system will
Atomic Entity interact with an individual entity
State Machine a direct graph
State circumstances of an entity
Transition any given time a state change
Attribute a named variable
Value intangible individual entity
Event individual taking place at some time
Physical Event event initiated by a context entity
• R and Rel are relations expressed in Table 2. It
provides the associations between ontology con-
cepts.
Table 2: Some Associations of Environment Ontology.
Association formation
has
−
static
−
property AtomicEntity→Attribute9Value
has
−
dynamic
−
property
Causellingntity→StateMachine
can
−
issue
AutonomousEntity→PhysiEvent
Causellingntity→PhysicalEvent
has
−
state
StateMachine→State
has
−
in
−
event
StateMachine→Event
has
−
transition
StateMachine→Transition
source
−
from
Transition→State
be
−
triggered
−
by
Transition→Event
cause
Transition→PhysicalEvent
cause event e ↑ event
anonymous cause event
ε ↑ event
cause event sends value
e ⇑ event(attr.val)
in state
e : state
send value
e ։ attr.val
receive value
e և attr.val
need value
e x e
′
.attr.val
• As for interaction associations, we use notation
set {↑, ↓, ⇑, ⇓, :, ։, և, x} to specify the relations
between an entity and its events to represent inter-
actions.
We use the following interaction events depen-
dency to capture the relations among state transitions.
• Task Dependency(TSD): If PostState(T
1
) =
PreState(T
2
), then T
2
task depends on T
1
, i.e.
T
2
T
T
1
.
• Selection Dependency(SLD): If PreState(T
1
) =
PreState(T
2
), then T
2
selection depends on T
1
,
i.e.T
2
S
T
1
.
• Outer Task Dependncy(OTSD): Let e
1
↑ event
1
represents entity
1
causes event
1
, e
2
↓ event
2
repre-
sents entity
2
accepts event
2
. If e
2
↓ event
2
should
happen after e
1
↑ event
1
, then e
1
↑ event
1
outer
task depends on e
2
↓ event
2
. i.e. e
1
↑ event
1
O
e
2
↓
event
2
.
2.2 Environment Ontology of OSD
Domain
Domain Entity is environment element in a specific
domain, it is depicted by five elements. DE :=<
Ename, TypeSet, StaticPropertySet, DynamicState
Machine, Interaction >
• Ename is entity’s name.
• TypeSet is the set of Types the entity belongs to.
Usually, an entity belong to more than one type.
• StaticPropertySet is the set of static property the
entity has.
• DynamicStateMachine depicts the dynamic prop-
erties and state transitions triggered by interac-
tions.
• Interaction describes an entity’s actions it causes
to happen or actions caused by other entities.
Figure 1(Jin and Liu, 2006) shows the environ-
ment of On-Line Selling Domain. Each entity is
expressed in a rectangle with 5 elements introduced
above. Table 3 gives some OTSD of On-Line Selling.
Table 3: OTSD of On-Line Selling Domain.
Customer ⇑ pickout(Product.Name.Value)
Product ⇓ BeingPickedOut(Product.Name.Value)
MoneyCollector ↑ HavePaid
Customer ↓ HavePaid
MoneyCollector ↑ HavePaid
Product ↓ ItemsBeenPaid
Customer ↑ Confirm
Product ↓ OrderBeenCon firmed
Customer ↑ Confirm
Order ⇓ Confirmed(TotalPrice.value)
3 WEB SERVICE DESCRIPTION
BASED ON ENVIRONMENT
ONTOLOGY
After describing the environment ontology
of OSD domain, we present a triplet to de-
fine capability description: CapabilityDes :=<
EnvironmentEntityS, InteractionS, Scenario >
• EnvironmentEntityS: EnvironmentEntity is the
set of environment entities of OSD domain which
are involved during the Web Service execu-
tion. EnvironmentEntityS := {entity|entity ∈
WEB SERVICE RETRIEVAL BASED ON ENVIRONMENT ONTOLOGY
383
Member
Anonymous
OffLine
#i#applyMemebrship
(Name,Address,Contact),
#r#CID.value
#i#withdraw
#i#logIn(CID),
#t#welcome
#i#logOut,
#t#byebye
Customer(Name, Address, Contact, CID)
Online
InPaying
Choosing
Null
#i#continue
#r#Product.<Name,Price>.value,
#i#select(Product.Name.value)
#i#confirm
#r#Product.<Name,Price>,
#i#select(Product.Name.value)
#i#cancel
sink
sink
C,A,S
sink
Confirmed
BeingFilled
Empt
y
Null
#t#create(OID,CID,_,0)
#t#confirmed(TotalPrice.value)
#t#addInItem(Product.<Name,Price>.value)
#t#reset
#t#cancel
#t#cancel
Order(OID,CID,Product*,TotalPrice)
#t#addInItem(Product.<Name,Price>.value)
C,S
go to payment-
related session
#i#orderBeenConfirmed
start vertex end vertex
C,S
Manager
#i#CancelMembership(CID)
A
child state
machine of
`Member'
child state
machine of
`Online'
Entity State l Transition
#t#CancelMembership(CID)
Start
#i#start
Paid
InShop
#i#enter
#i#handOver(Product.Name.value),
#t#havePaid
#i#goOut,
#t#byebye
Product(Name,Price)
InStock
ByPickout
BeingSold
#t#beingPickedOut
#t#orderBeenConfirmed
#t#itemsBeenPaid
MoneyCollector
#i#collectingMoney(Amount)
#i#havepaid
A
#r#Product.<Name,Price>,
#i#pickout(Product.Name.value)
go to payment-
related session
Figure 1: Environment of On-Line Selling Domain.
OSDEntity},in which OSDEntity =
{Product, MoneyCollecter, Manager,Customer,
Order}.
• InteractionS: InteractionS contains the
environment entitys possible interactions.
InteractionS :=
Entity
i
∈ContextEntityS
Entity
i
∗ EI.
• Scenario: A scenario element depicts a in-
tegral interaction process which may cause
several state transitions represented by Effect.
Scenario := {InteractionSequencekE f fect}.
IntercationSequence is the set of sequen-
tial Event interactions, which belongs
to several environment entities. And
E f fect := {CE : CE.State
1
→ CE.State
2
}.
4 WEB SERVICE DISCOVERY
We propose that Web Service request annotation
mechanism is the same with Web Service description
profile. Now suppose we have a request for a Web
Service with this effect:
Product : InStock → BeingSold
The Web Service discovery process works as fol-
lows.
Step 1.Capture the request into Web Service descrip-
tion profile.
1
Extract the state transition caused by the re-
quired Web Service. Find the path which can proceed
from start state to end state in context entity’s Dynam-
icStateMachine in the guide of Web Service requester.
Get the sequence of event interactions along this path:
< INT
1
, INT
2
, ··· , INT
n
>.
2
Check the dependency of the sequential set el-
ements one by one to get the entire interaction se-
quence. The algorithm of finding all OTSD and TSD
dependencies of an INT works recursively as follows:
Procedure LocatInteract(INT)
If INT prefix=’#i#’
Then find post interaction of INT
INTopo in OTSD;
Insert it after INT;
LocatInteract(INTopo);
Find pre and post interaction of INT
INTtpr,INTtpo in TSD;
Insert them before and after INT;
LocatInteract(INTtpr);
LocatInteract(INTtpo);
End Then
End If
Find Pre interaction of INT INTopr in OTSD;
Insert it before INT;
LocatInteract(INTopr);
Find pre and post Interaction of INT
INTtpr,INTtpo in TSD;
Insert them before and after INT;
LocatInteract(INTtpr);
LocatInteract(INTtpo);
End Procedure
3
Construct the request description. Get the re-
lated entities when refer to OTSD in step1.2. In step
1.2 it is possible that more than one sequential sets
are approved by the requester, which with the effect,
the state transition, construct the scenarios. When all
these finished, the Web Service request has the form
of:
< EnvironmentEntityS
re
, InteractionS
re
, Scenario
re
>
And the scenario part of request description is showed
in table 4.
Step 2.Retrieve the Web Service
1
Consider the standby Web Services in Web Ser-
vice registration center. Every Web Service has the
description of:
< EnvironmentEntityS
sd
, InteractionS
sd
, Scenario
sd
>
WEBIST 2007 - International Conference on Web Information Systems and Technologies
384
Select f rom Standby Web Service where
EnvironmentEntityS
re
⊂ EnvironmentEntityS
sd
And InteractionS
re
⊂ InteractionS
sd
And
Scenario
re
⊂ Scenario
sd
With this process, the initial request is for the op-
tion ”a selling service by order”, and the proper Web
Service can be retrieved.
Table 4: Profile of Selling by Order.
OrderSeller
Scenario:{
{Customer(CID)↑start,
Order⇓create(OID,CID,
/
0,0),
CustomerxProduct.<Name,Price>.vlaue*,
{Customer⇑select(Product.Name.value),
Order⇓addInItem(Product.<Name,Price>.value)}*,
Customer↑confirm,
Order⇓confirm(TotalPrice.value),
Customer↑continue}
kProduct.Name:InStock Product.Name:
BeingSold,
{Customer(CID)↑start,
Order⇓create(OID,CID,
/
0,0),
CustomerxProduct.<Name,Price>.vlaue*,
{Customer⇑select(Product.Name.value),
Order⇓addInItem(Product.<Name,Price>.value)}*,
Customer↑cancel},
kNull,
·· · }
5 RELATED WORK AND
CONCLUSION
WSDL uses syntax ports, OWL-S tries simple se-
mantic pre and post interface of a Web Service pro-
cess, both method are not semantic enough for auto-
matic feature of Web Services. Different from above
work, we construct a domain environment ontology
to support Web Service semantic description. Web
Service’s capability is grounded onto the state transi-
tions of environment entities, and the event interac-
tions among the entities. Based on the environment
ontology, we design a matching process for automatic
Web Service discovery and selection in Web Service
description framework. Also, this ontology-based de-
scription framework support dynamic composition of
Web Services by combining interaction sequence and
scenarios.
However, many issues still need consideration in
future. We are trying to use process algebra to de-
scribe the interaction of Web Services, representing
the capability of Web Services more potently.
ACKNOWLEDGEMENTS
This work was supported by the National Natural
Science Fund for Distinguished Young Scholars of
China under Grant No. 60625204, the Key Project
of National Natural Science Foundation of China un-
der Grant No.60496324, the National 973 Funda-
mental Research and Development Program of China
under Grant No.2002CB312004, the National 863
High-tech Project of China, the Knowledge Innova-
tion Program of the Chinese Academy of Sciences
and MADIS.
REFERENCES
Berners-Lee, T., Hendler, J., and Lassila, O. (2001). The
Semantic Web. Scientific American, America.
de Vergara, J. E. L., Villagr, V. A., and Berrocal, J. (2005).
Application of owl-s to define management interfaces
based on web services. Management of Multimedia
Networks and Services, 3754/2005:242–253.
E.Christensen, F.Curbera, G.Meredith, and S.Weerawarana
(2001). Web Services Description Language (WSDL)
1.1. Technical Report NOTE-wsdl-20010315, W3C.
http://www.w3.org/TR/2001/NOTE-wsdl-20010315.
Fu, X., Bultan, T., and Su, J. (2004). Analysis of interacting
bpel web services. In Proceeding of 13th World Wide
Web Conference, pages 621–630, New York, USA.
Gutierrez, C., Hurtado, C., and Vaisman, A. (2005). The
Semantic Web: Research and Applications, volume
3532/2005. Springer Berlin Heidelberg.
Jin, Z. and Liu, L. (2006). Web service retrieval, an ap-
proach based on context ontology. In Proceedings of
the 30th Annual International Computer Software and
Applications Conference, pages 513–520, Chicago,
USA.
M.Jackson (1997). The meaning of requiremens. Annals of
Software Engineering, 3(5):5–21.
M.Jackson (2001). Problem Frames: Analyzing and struc-
turing software development problems. Addison-
Welsley.
Motta, E., Domingue, J., Cabral, L., and Gaspari, M.
(2003). Irscii: A framework and infrastructure for
semantic web services. Proceedings of 2nd Inter-
national Semantic Web Conference, 2870/2003:306–
318.
Paolucci, M. and Sycara, K. (2003). Autonomous semantic
web services. IEEE Internet Computing, 7(5):34–41.
WEB SERVICE RETRIEVAL BASED ON ENVIRONMENT ONTOLOGY
385
