A COMPARATIVE STUDY ON SEMANTIC WEB SERVICES

FRAMEWORKS FROM THE DYNAMIC ORCHESTRATION

PERSPECTIVE

Domenico Redavid, Floriana Esposito

Computer Science Dept., University of Bari, Via Orabona 4, 70126, Bari, Italy

Luigi Iannone

Computer Science Dept., University of Manchester, Kilburn Building, Oxford Road, M13 9PL, Manchester, U.K.

Keywords:

Semantic web services, Dynamic orchestration.

Abstract:

This paper presents a comparison between the two main existing frameworks for modelling Semantic Web

Services, namely OWL-S and WSMO, based on the formal support they provide for reasoning. The adopted

yardstick is the set of use cases dictated by one of the foremost tasks in the research ﬁeld: the dynamic

orchestration. As explained in the paper, the term dynamic denotes the capability of an agent to design and

manage in an automatic way an orchestration schema using the semantic descriptions of some services. This

capability, strictly related to the automatization of Discovery, Selection, Composition and Invocation use cases,

is constrained by the knowledge representation strategies adopted by WSMO and OWL-S. As consequence,

each approach has some limits discussed in this work.

1 INTRODUCTION

From the Web services perspective, an orchestration

is a declarative speciﬁcation that describes a work-

ﬂow supporting the execution of a speciﬁc business

process, operation, or service (Peltz, 2003). Cur-

rently, Web services technologies make it possible

to describe orchestration, but only at design-time.

Semantic Web Services (SWS) provide an ontologi-

cal framework for describing services in a machine-

readable format. In (McIlraith et al., 2001) a software

agent applies logical reasoning on SWS descriptions

in order to provide on the ﬂy orchestration capabil-

ities. With the adjective dynamic we denote the ca-

pability of an agent to design and manage in an auto-

matic way an orchestrationschema using the semantic

descriptions of some services. The notion of dynamic

orchestration becomes useful in scenarios where there

is the need of run-time designing of process integra-

tion using semantic descriptions of the entities in play.

This work aims at answering to the following ques-

tions:

• What does orchestrating mean for the Semantic

Web?

• What are the requirements to enable the dynamic

SWS orchestration?

• How well does each examined framework support

such requirements?

In the Section 2 we revisit the notion of SWS orches-

tration taking into account the semantics of services.

In Sections 3 we brieﬂy describe and analyze the sup-

port offered to it by the two leading efforts for SWS

representation (OWL-S and WSMO)

. In particular,

we examine the formal support they provide for rea-

soning on the semantic descriptions of the services

and the state of the art. In Section 4 a discussion on

the limits of each approach is presented.

2 ORCHESTRATION AND SWS

INFRASTRUCTURE

Orchestration for Web services has to rely on XML-

based service description. Their merely syntactical

Both the leading Semantic Web Services efforts

- OWL-S (http://www.w3.org/Submission/OWL-S) and

WSMO (http://www.w3.org/Submission/WSMO), are cur-

rently submissions in the Semantic Web Services Standard-

ization process.

355

Redavid D., Esposito F. and Iannone L..

A COMPARATIVE STUDY ON SEMANTIC WEB SERVICES FRAMEWORKS FROM THE DYNAMIC ORCHESTRATION PERSPECTIVE.

DOI: 10.5220/0003071003550359

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2010), pages 355-359

ISBN: 978-989-8425-29-4

 2010 SCITEPRESS (Science and Technology Publications, Lda.)

nature (Peltz, 2003) represents an unsurmountable

obstacle towards the automatic implementation of the

necessary operations to accomplish dynamicity in or-

chestration. Let us, indeed, consider the following

scenario:

“Given a request (goal), an agent (dynamic orchestra-

tor) discovers the possible SWSs able to accomplish

the goal. At the same time, it composes the discovered

services in order to have more possible ways to reach

the goal. Having more choices available, it selects

the best one exploring functional and non-functional

properties of the different SWSs. The selection pro-

cess can be used also during the composition (sub-

goal matching). Finally, the same agent manages the

correct invocation of the selected services.”

In this paper, dynamic orchestration of SWS

means carrying out one or more of the operation men-

tioned above. This is very different from the de-

sign of the work-ﬂow of the execution of simple Web

services (Peltz, 2003). The ﬁrst step to realize this

scenario is the automatization of the emphasized op-

erations (i.e.: discover, composition, selection and

invocation). These operations, amongsts the others

(namely, publishing, deployment and ontology man-

agement which are not directly involved in our sce-

nario), are already been identiﬁed as use cases for the

SWS infrastructures into the Usage Activities dimen-

sion described in Figure 4 in (Cabral et al., 2004). In

the same work the architectural components and se-

mantic descriptions needed to realize the Usage Ac-

tivities are been individuated. The former had been

grouped under the Architecture dimension and in-

clude a register, a reasoner, a matchmaker, a decom-

poser, and an invoker. The latter, make up the Service

Ontology dimension and speciﬁes the semantics of:

• Functional capabilities, such as inputs, output,

pre-conditions and post-conditions;

• Non-Functional capabilities, such as category,

cost and quality of service;

• Provider related information, such as company

name and address;

• Task or goal-related information and domain

knowledge deﬁning, for instance, the type of the

inputs of the service.

The dynamic SWS orchestration, as we deﬁned

above, needs to be matched with these dimensions

(reported in Table 1) in order to individuate its re-

quirements.

In the architecture dimension, the reasoner is the

most important component to implement these use

cases because it allows the semantic matchmaking of

Table 1: Orchestration in the SWS Infrastructure.

DIMENSION DYNAMIC ORCHESTRATION

Usage Composition, Selection,

Activities Invocation, Discovery

Architecture (De-)Composer, Reasoner,

Invoker, Matchmaker

Service Inputs, Outputs,

Ontology Conditions,

Atomic/Composite services

the services properties enabling the automatic (de-

)composition of the services, and the automatic choice

of the right parameters for the services invocation.

Service ontology dimension is necessary because it

makes possible to deﬁne the semantics of functional

and non-functional properties owned by atomic and

composite services that participate to the orchestra-

tion, as well as the semantics of their control con-

structs and data-ﬂow. Ideally, the semantic descrip-

tions of inputs and outputs should exist independently

from the SWS speciﬁcation. They should be cho-

sen within existing ontologies describing a particu-

lar knowledge domain and, at most, reﬁned. Hence,

in order to deﬁne atomic SWS the semantic repre-

sentation of conditions over inputs and outputs is

needed, whereas to deﬁne composite SWS the se-

mantic representation of control constructs and data-

ﬂow is needed too. Dynamic orchestration becomes,

than, mainly a representation problem. This is why,

in the following, when we will compare and contrast

WSMO and OWL-S we will focus on the knowledge

representation infrastructure.

3 WSMO AND OWLS

The Web Service Modeling Ontology (WSMO) (Ro-

man et al., 2005) provides a framework for semantic

descriptions of Web services. It consists of four core

elements that, properly linked, constitute the semantic

description of the service: Ontologies, Goals, Web

Services, and Mediators. The semantics of these en-

tities can be speciﬁed using one of the formal lan-

guages deﬁned by the Web Service Modeling Lan-

guage

(WSML). WSML includes several language

variants, based on three different logical formalisms,

namely, Description Logics (Baader et al., 2003)

(WSML-DL), First-Order Logic (Enderton, 1972)

(WSML-Full) and Logic Programming (Lloyd, 1987)

(WSML-Rule and WSML-Flight). Furthermore, it

Web Service Modeling Language (WSML),

W3C Member Submission, 3 June 2005.

http://www.w3.org/Submission/WSML/

KEOD 2010 - International Conference on Knowledge Engineering and Ontology Development

356

deﬁnes the variant WSML-Core that can be identiﬁed

as the intersection between a particular Description

Logic (a subset of S H I Q ) and Horn Logic (without

function symbols and equality) (Grosof et al., 2003).

WSML DL is, in general, incompatible with both

WSML-Flight and WSML-Rule. Therefore, com-

plete reasoning about services descriptions speciﬁed

using WSML DL, Rule and Flight is unfeasible be-

cause WSML Full, the common super-language, is

undecidable.

The Web Ontology Language for Services (OWL-

S) is an OWL

ontology describing the three essen-

tial aspects of a service (i.e. its adervetisment, pro-

cess model, and protocol details) using three differ-

ent modules: Service Proﬁle, Service Model, and

Service Grounding. OWL-S itself is an OWL on-

tology. OWL, whose formal foundation is the De-

scription Logics (Baader et al., 2003), provides three

increasingly expressive sub-languages, in detail:

• OWL-Lite can be used to model classiﬁcation

hierarchies and simple constraints. OWL-Lite

has the lowest computational complexity amongst

OWL sub-languages.

• OWL-DL is used when the maximum decid-

able expressivity is required. It corresponds to

S H O I N (D ) DL (Horrocks et al., 2003).

• OWL-Full, although allowing for as much ex-

pressivity as RDF, is undecidable therefore it

could be hardly be used for reasoning.

However, as pointed out in the following, SWRL (in

its decidable fragment) represents a reasonable ad-

vancement in reconciling Logic Programming and

Description Logics. For this reason, OWL-S is in a

more better position than WSMO as SWRL can be

seamlessly integrated with OWL.

The state of the art, both for WSMO and OWL-S, can

be breﬂy analyzed w.r.t. the use cases reported in the

usage activities dimension of Table 1, i.e. Discov-

ery, Selection, Composition, and Invocation. Please,

notice that, whilst for WSMO exist different imple-

mented infrastructures that unify, amongst the oth-

ers, all the aspects described above, to the best of

our knowledge there is not any counterpart developed

for OWL-S, yet. For WSMO, in general, the com-

mon starting point for the implementation of such

functionalities is either WSMX

or IRS-3

. (Kifer

OWL-S is based on OWL 1.0 recently extended

with the new recommendation OWL 2, as reported in

http://www.w3.org/TR/owl-overview/

Web Service Execution Environment (WSMX) -

http://www.wsmx.org

Internet Reasoning Service (IRS) 3 -

http://kmi.open.ac.uk/technologies/irs/

et al., 2004) presents a logical framework that exploits

WSMO formal descriptions to dynamically discover

Web Services matching the requester goals. The SWS

discovery in OWL-S, in most cases, aims to enhance

the UDDI registry and to exploit semantic matchmak-

ing techniques applied to the service proﬁle model.

Lynch et al. (Lynch et al., 2006) explore a potential al-

ternativefor UDDI registries

, and investigatethe fea-

sibility of a general publish/subscribe model for Web

Service Discovery. The research on services selection

in WSMO is often related to the service discovery be-

cause both use cases require a semantic matchmaker

component. The selection mechanism of OWL-S is

related to the application of semantic matchmaking

techniques both to the service proﬁle and to the pro-

cess model. The invocation use case is strictly related

to the semantics of the protocol used to invoke the ser-

vice. In (Walton, 2005) a particular protocol (Which

can be used with OWL-S and WSMO) to represent

the computational aspects of a Web service, such as

invocation order, and inter-argument dependencies, is

proposed. The research on composition of services

encompassesmore elements within service ontologies

than other usage activities because it must consider

the SWS conditions during the composition task. To

the best of our knowledge, works on fully automatic

composition of WSMO services have not been pro-

posed yet. However, a semi-automatic composition

approach is proposed in (Hakimpour et al., 2005) .

Most OWL-S research activities focused on compo-

sition methods drawing inspiration mainly from AI

planning, like reported in (McIlraith and Son, 2002),

one of the ﬁrst works about this topic. In (Redavid

et al., 2008) a method for encoding OWL-S atomic

processes by means of SWRL rules and composing

them using a backward search planning algorithm is

presented. The proposed solution has been realized

using exclusively Semantic Web technologies.

4 DISCUSSION

The meaning of orchestration once we take semantics

into account becomes different from the pre-arranged

Web service composition, which is constrained only

at syntactic level. To summarize, to orchestrate means

to realize the automatic discovery, selection, compo-

sition, and invocation of services so as to achieve a

goal. Solving this problem requires to respond to use

cases organized along three dimensions in the SWS

infrastructure (please see Sec. 2). Here we focus on

the usage activities and service ontology to compare

Universal Description Discovery and Integration

(UDDI) - http://uddi.xml.org/uddi-org

A COMPARATIVE STUDY ON SEMANTICWEB SERVICES FRAMEWORKS FROM THE DYNAMIC

ORCHESTRATION PERSPECTIVE

357

and contrast the two main framework proposed so far:

WSMO and OWL-S

With respect to the Usage activities dimension,

both WSMO and OWL-S take into account all the use

cases that we have individuated for the dynamic or-

chestration. They differ only in the fact that WSMO

requires the modeling of the service requester fea-

tures. This implies the need for tackling the interoper-

ability problems between different WSMO elements

within the framework itself. In order to overcome this

issue, different types of mediators are been deﬁned

and considered in every usage activity. This is re-

ﬂected also in the Service ontology dimension where

WSMO presents a richer service description model

than OWL-S. Consequently, some extra features re-

lated to Web services execution (like goals, mediator

and communication protocols) are represented. The

shortcoming of this approach is that any WSMO ser-

vices usage is limited to its declared goals at design

time.

The formal semantics offered by WSMO and

OWL-S is fundamental to achieve dynamic orches-

tration. In Figure 1 we compare the formalisms on

which both frameworks are based.

WSMO appears unrelated to the Semantic Web

stack of languages and formalisms, with the excep-

tion of its basic level (URI and XML) that guarantees

syntactic interoperability on the Web. It offers map-

pings with the RDF syntax and the S H I Q Descrip-

tion Logic which, however, does not cover the whole

OWL-DL.

WSMO global approach to service ontology deﬁ-

nition can be summarized as follows:

• It started with the deﬁnition of a FOL-like lan-

guage able to represent of all the possible aspects

of a Web service;

• It deﬁned several sub-languages restricting the

original expressive power to well-known frag-

ments (DL or LP);

• It (partially) mapped such fragments on to SW

current standard languages.

Given the known incompatibility between DL and LP

(Borgida, 1996) currently the only implemented rea-

soning available for WSMO, when dealing with both

rules and ontologies, is the one restricted to their com-

mon subfragment, i.e. DLP. Hence WSMO inference

capabilities, as far as the implementation goal, to the

best of our knowledge are signiﬁcantly reduced with

Semantic Annotations for WSDL and XML

Schema (SAWDL) - W3C Recommendation,

http://www.w3.org/TR/sawsdl/- is not been considered

because it does not allows the process model representa-

tion.

respect to a pure Semantic Web based counterpart.

OWL-S, instead, is based natively in OWL whose

natural extension towards rules is SWRL. SWRL de-

cidable fragment (DL-safe rules (Motik et al., 2005))

is the largest possible union of primitives from both

formalisms rather than their intersection. Current

improvements of DL reasoners, now able to handle

SWRL rules, allowed to developearly prototypes able

to compose SWS described using OWL-S process

model and mapped into SWRL (Redavid et al., 2008).

Therefore, from the perspective of two out of

three main dimension of SWS infrastructures, OWL-

S emerges has better equipped to encompass all the

requirements dictated.

5 CONCLUSIONS

The SWS frameworks proposed in literature provide

different support to dynamic orchestration. In this

work we conducted a comparative study to elicit dif-

ferences and analogies to do that, and remarked the

capabilities necessary to enable dynamic orchestra-

tion: the needed requirements and the suitability of

OWL-S and WSMO to support such requirements. As

showed in the Section 2, the set of requirements are

implied by means of the use cases, namely automatic

discover, selection, composition and invocation, re-

quired to make dynamic the SWS orchestration. The

formal language underlying the SWS frameworks is

the key for an effective realization of these use cases.

For this reason, in Sections 3 we described the formal

support enabling reasoning on the semantic descrip-

tions of the services offered by WSMO and OWL-

S (based on OWL+SWRL and WSML, respectively).

Finally, in Section 4, we have compared the for-

malisms underlying OWL+SWRL and WSML from

the point of view of the expressive power effectively

exploitable for reasoning on service descriptions. As

result of this comparison, OWL-S emerges as more

suitable candidates for the dynamic orchestration. As

future work we will focus on two aspects. First, deal-

ing with the Semantic Web natively issues, for in-

stance, the absence of primitives for retracting knowl-

edge due to the monotonic nature of a DL Knowledge

base and the semantic interoperability. Second, the

realization of the software agents able to manage the

dynamic SWS orchestration by considering low-level

details as, for instance, Quality of Service, Service

Level Agreement and the coordination for the con-

crete Web services Invocation.

KEOD 2010 - International Conference on Knowledge Engineering and Ontology Development

358

Figure 1: A comparison between the formal support provided by OWL+SWRL and WSML.

REFERENCES

Baader, F., Calvanese, D., McGuinness, D., Nardi, D., and

Patel-Schneider, P., editors (2003). The Description

Logic Handbook. Cambridge University Press.

Borgida, A. (1996). On the relative expressiveness of de-

scription logics and predicate logics. Artiﬁcial Intelli-

gence, 82(1-2):353–367.

Cabral, L., Domingue, J., Motta, E., Payne, T. R., and

Hakimpour, F. (2004). Approaches to Semantic Web

Services: an Overview and Comparisons. In Bus-

sler, C., Davies, J., Fensel, D., and Studer, R., edi-

tors, ESWS, volume 3053 of Lecture Notes in Com-

puter Science, pages 225–239. Springer.

Enderton, H. B. (1972). A Mathematical Introduction to

Logic. Academic Press, New York.

Grosof, B. N., Horrocks, I., Volz, R., and Decker, S. (2003).

Description logic programs: combining logic pro-

grams with description logic. In Proceedings of the

12th international conference on World Wide Web,

pages 48–57, New York, NY, USA. ACM Press.

Hakimpour, F., Sell, D., Cabral, L., Domingue, J., and

Motta, E. (2005). Semantic web service composition

in irs-iii: The structured approach. cec, 00:484–487.

Horrocks, I., Patel-Schneider, P. F., and van Harmelen, F.

(2003). From SHIQ and RDF to OWL: the making of

a Web Ontology Language. J. Web Sem., 1(1):7–26.

Kifer, M., Lara, R., Polleres, A., Zhao, C., Keller, U.,

Lausen, H., and Fensel, D. (2004). A logical frame-

work for web service discovery. In In Proc. of the

ISWC 2004 workshop on Semantic Web Services:

Preparing to Meet the World of Business Applications.

Lloyd, J. W. (1987). Foundations of Logic Programming,

2nd Edition. Springer-Verlag.

Lynch, D., Keeney, J., Lewis, D., and O’Sullivan, D. (2006).

A Proactive approach to Semantically Oriented Ser-

vice Discovery. In Proceedings of the Second Work-

shop on Innovations in Web Infrastructure (IWI 2006),

Edinburgh, Scotland.

McIlraith, S. A. and Son, T. C. (2002). Adapting golog

for composition of semantic web services. In Fensel,

D., Giunchiglia, F., McGuinness, D. L., and Williams,

M.-A., editors, KR, pages 482–496. Morgan Kauf-

mann.

McIlraith, S. A., Son, T. C., and Zeng, H. (2001). Semantic

Web Services. IEEE Intelligent Systems, 16(2):46–53.

Motik, B., Sattler, U., and Studer, R. (2005). Query answer-

ing for owl-dl with rules. Journal of Web Semantics:

Science, Services and Agents on the World Wide Web,

3(1):41–60.

Peltz, C. (2003). Web Services Orchestration and Choreog-

raphy. Computer, 36(10):46–52.

Redavid, D., Iannone, L., Payne, T. R., and Semeraro, G.

(2008). OWL-S Atomic Services Composition with

SWRL Rules. In An, A., Matwin, S., Ras, Z. W., and

Slezak, D., editors, ISMIS, volume 4994 of Lecture

Notes in Computer Science, pages 605–611. Springer.

Roman, D., Lausen, H., de Bruijn, J., Lara, R., Stollberg,

M., and Polleres, A. (2005). Web Services Modeling

Ontology. Journal of Applied Ontology, (1)1, 77-106,

2005.

Walton, C. (2005). Protocols for web service invocation. In

Proceedings of the AAAI Fall Symposium on Agents

and the Semantic Web (ASW05).

A COMPARATIVE STUDY ON SEMANTICWEB SERVICES FRAMEWORKS FROM THE DYNAMIC

ORCHESTRATION PERSPECTIVE

359