Professional Documents
Culture Documents
Me Cheri 2019
Me Cheri 2019
3, 2019 209
Karima Mecheri*
LISCO Laboratory, Computer Science Department,
Badji Mokhtar University,
Annaba, Algeria
Email: mecherika@yahoo.fr
*Corresponding author
Mahmoud Boufaida
LIRE Laboratory,
Abdelhamid Mehri – Constantine 2 University,
Constantine, Algeria
Email: mahmoud.boufaida@univ-constantine2.dz
Keywords: SWS; semantic web services; interoperability; semantic model; abstract state
machine; behavioural context; mediation; ontologies.
Reference to this paper should be made as follows: Mecheri, K., Boufaida, M., Meslati, D.
and Souici-Meslati, L. (2019) ‘Context-based interoperability of semantic web services’,
Int. J. Metadata, Semantics and Ontologies, Vol. 13, No. 3, pp.209–226.
Mahmoud Boufaida is a Full Professor in the Computer Science Department of the University of
Constantine 2, Algeria. He Heads the research group Information Systems and Knowledge Bases.
He has published several papers in international conferences and journals. He has managed and
initiated multiple national and international level projects including interoperability of
information systems and integration of applications in organisations. He has been program
committee member of several conferences. His research interests include cooperative information
systems, web databases and software engineering.
Djamel Meslati is a Full Professor in the Computer Science Department of Badji Mokhtar
University, Annaba, Algeria. He Heads the LISCO Laboratory. He published several papers in
international conferences and journals. He has been program committee member of several
conferences. His research interests include bio-inspired systems and software engineering.
Labiba Souici-Meslati is a Full Professor in the Computer Science Department of Badji Mokhtar
University, Annaba, Algeria. She Heads a research team in LISCO Laboratory. She has published
several papers in international conferences and journals. Her research interests include machine
learning, bio-inspired systems and application of artificial intelligence techniques in software
engineering.
Heterogeneities between non-functional properties: All 2.2 Semantic description of web services
the properties that characterise a web service and that are
not directly linked to the functionality delivered by the We distinguish mainly two approaches for describing the
service, are considered non-functional. The non-functional semantics of web services (Mrissa, 2007): (1) the approaches
properties are not considered in the WSDL standard exploiting semantic description languages (OWL-S (Martin
et al., 2004; Sikos, 2015), DE, DSD (Klein et al., 2005), as well
language. Their heterogeneities include (Mrissa, 2007):
as the conceptual architecture WSMO (Arroyo and Stollberg,
The ACID transactional properties. 2004)) and (2) The annotation-based approaches of existing
The Quality of Service (QoS) (Pengcheng et al., 2018) languages. These annotations may concern:
and performances of web services such as availability, Business processes such as SESMA (SEmantic Service
velocity, cost, reliability, and so on. MArkup) (Peer, 2005).
Heterogeneities between functional properties: They concern The registers UDDI and ebXML, using the languages
the properties of the services usually described in a like DAML-S (Paolucci et al., 2002; Dogac et al., 2004)
WSDL document such as input/output parameters, service and SAWSDL (Hausberger, 2009).
functionality, exchange protocol used and encoding.
Heterogeneity of the exchanged data: Data exchanged The WSDL description language either by exploiting
between web services can present heterogeneities that can the extensibility elements of the WSDL with the
be classified in three levels: WSDL-S (Sheth et al., 2008; Miller et al., 2004) or by
using the SAWSDL (Hausberger, 2009; Midouni et al.,
The syntactic level: it concerns the encoding of data. The
2016).
heterogeneities of this level are dealt with using XML.
Semantically annotating web services is gaining lots of
The structural level: it is relative to the different
attention as a significant area to support the machine-driven
representations of data to the level schema.
matchmaking and composition of web services. Therefore,
The semantic level: it includes the significance conveyed the support of ontologies and tools for the semantic
by data. annotation of web services is turning into a key concern to
assist the dissemination of SWS (Shridevi and Raju, 2018).
Heterogeneities between behavioural properties: The
communication between the WS requester and the WS
provider may take place only if their Message Exchange 2.3 Integration approaches of web services
Patterns (MEPs) match precisely. That is, when the first sends a For the integration approaches of the web services, Mrissa
message, the second is able to receive it. Heterogeneity of (2007) distinguished the following categories of approaches:
MEPs is very complex than that of data. Fensel and Bussler
(2002) identify three cases of mismatches which may appear Approaches based on adapters (Medjahed et al., 2004).
during message exchange (choreography): precise match, Community-based approaches (Taher et al., 2006;
resolvable message mismatch and irresolvable message Metrouh and Mokhati, 2018).
mismatch. The second case is related to:
Approaches based on extended descriptions (Dumas
the presence of additional messages, and Spork, 2006).
the different order of messages, Web services-based approaches: These approaches rely
the need to splitting a message, on web services for publishing, invocation, mediation,
and execution.
the need to combining messages,
The latter two approaches are more promising compared to
and the lack of required messages. the former because they have the advantages of service-
Cimpian and Mocan (2005) identify five process mediation oriented architectures (Izza et al., 2008; Domingue et al.,
patterns: 2008; Haller et al., 2005; Cimpian and Mocan, 2005; Vitvar
et al., 2009; Khater et al., 2017).
stopping an unexpected message,
inversing the order of messages, 2.4 Mediation of web services
splitting a message, The mediation is an aspect of composition for resolving the
combining messages and conflicts caused by the heterogeneities that can appear
between web services, to ensure their interoperability.
generating a message.
In the literature, there are several categories of mediation
In order to solve the problem of heterogeneities of web between web services, because it is possible to classify the
services and to ensure their integration, we need firstly, a mediation tasks according to various perspectives.
semantic description of these services using ontologies, Bussler (2003) distinguished two types of mediation: the
which gives us SWS, and secondly, a mediation of these business event mediation and the business process one. The
services. Then, the mediation is dynamic and is achieved first aims to resolve data heterogeneities while the second
due to the semantic description. aims to resolve behavioural ones.
212 K. Mecheri et al.
Cabral and Domingue (2005, 2012) proposed three levels: Data mediation level: solving data conflicts that can
data mediation, functionality mediation, and business occur between web services.
process one.
Functional mediation level: establishing a correspondence
Mrissa (2007) proposed three mediation levels:
between the provided functionality and the requested one.
Web services integration level: it aims to solve all the
Process mediation level: solving behavioural heterogeneities
heterogeneities between the web services, notably those
that can occur between the two choreographies of web
between the properties not described in the WSDL
services in interaction.
documents (non-functional), as the properties concerning
the categorisation, the QoS, the sequences of exchange, Non-functional mediation level: establishing a
etc. correspondence between the non-functional properties of
the web service provider and those of the web service
Interface adaptation level: it aims to solve the
requester.
heterogeneities between the functional properties
described in the WSDL documents, as functionalities The most important parts of the interoperability process
offered by a web service, the protocols and the used are the semantic model and the mediation. They will be
encoding (RPC, Literal Document.). described after classification of SWS interoperability works.
In a previous work, we have presented the data mediation
Data mediation level: it aims to solve the heterogeneities
level (Mecheri et al., 2012). This paper is dedicated to the
of the data exchanged between web services. This level is
business process mediation level.
a sub-level of the interface adaptation that is itself a sub-
level of the web services integration.
The three classifications do not deal with the different 3 Classification and comparison of SWS
aspects of web services as security, QoS, and so on. The interoperability works
classification of Mrissa (2007) distinguished the functional
aspects and the non-functional ones of web services, but he In this section, we propose a classification and a comparison of
merges the behavioural aspect with the non-functional one. semantic interoperability works using web services based on
Our approach provides a flexible solution to make criteria we have selected (Mecheri and Souici-Meslati, 2010).
enterprise SWS interoperable at four levels: data, functional, We distinguish two main types of work: those that are based on
process, and non-functional. To this respect, our classification semantic languages belonging to the semantic web, and those
of web service mediation is based on four levels corresponding that are oriented towards semantic annotations of web service’s
to the following characteristics: existing languages (see Figure 1).
Figure 1 Classification based on web service description approaches
3.1 Approaches based on semantic languages as finite automata. However, the approach doesn’t deal with
heterogeneities.
The semantic service oriented architecture Web Service
Fonou-Dombeu and Huisman (2015) proposed an
Modelling eXecution (WSMX) environment (Haller et al.,
infrastructure for e-government SWS integrating concepts of
2005; DERI and STI2, 2017) is an environment of web
Business Process Modelling (BPM) and of semantic
services execution based on Web Services Modelling
annotations as tools for modelling and technological aspects of
Ontology (WSMO). This architecture allows the dynamic
SWS engineering for non-automated government operations
discovery, the invocation and the composition of the SWS.
and processes. The proposed infrastructure includes, among
WSMX ensures the interoperability of the Business to
others, three types of ontologies: domain, user, and service. It
Business (B2B) systems. This architecture is used by Vitvar
operates the WSMO and can be implemented via IRS-III and
et al. (2009) to present the DERI’s solution for solving SWS
WSMX.
challenge mediation scenario but these authors don’t give
In the behaviour approach for composite OWL-S
semantics to the message interaction. So, their algorithm
services discovery (Khater, 2017), the authors consider the
fulfils only the pattern “stopping an unexpected message”
matchmaking based only on service profile (inputs/outputs) not
and “combining messages” proposed by Cimpian and
sufficient in the case of composite services. To overcome
Mocan (2005) and given previously in page 2.
profile matching limitations, they propose an approach-based
The IRS-III architecture (Internet Reasoning Service)
sub-graph isomorphism to check the matching of process
(Cabral and Domingue, 2005; Domingue et al., 2008;
structure between two OWL-S process models of composite
Cabral and Domingue, 2012) is a framework permitting the
services and shortest path (OWLS-SP) to compute the score of
publication, the configuration and the execution of the
matching between nodes of service and those of query (atomic
heterogeneous web services. It is also based on the WSMO
services).
and contains the following components: Server, Publisher,
and Customer who communicate through the SOAP
protocol. IRS-III implements the data, the objective and the 3.2 Approaches based on semantic annotations
process mediations (Sikos, 2015). A process mediator is Mrissa (2007) proposed a context oriented semantic
used to solve the communication conflicts exploiting a set mediation architecture for the composition of web services.
of logical rules. The IRS-III specification lacks a necessary It is an architecture containing a provider layer, a
mechanism, techniques or an affiliation used to overcome composition layer and a description one. It deals with a data
heterogeneity (Arroyo and Sicilia, 2010). mediation using a mediator web service inserted between
Izza et al. (2008) proposed a service oriented architecture the sender web service and the receiver web service during
for integrating industrial information systems applications. This the composition execution. These services are previously
architecture, named Ontology-Driven Service-Oriented semantically described using the WSDL extensibility
Integration Architecture (ODSOIA), is an intra-enterprise elements and a domain ontology to which is associated the
solution that enlarges the Enterprise Service Bus (ESB) with contextual ontologies. However, the approach deals only
two architectural levels in addition to the service oriented with data mediation.
layer: the semantic level and the integration one. Different Kourtesis and Paraskakis (2008) presented the
mediations, several types of ontologies and mappings are implementation of the FUSION semantic registry, a
used. The authors don’t deal with the resolution of the data semantically-enhanced service registry that builds on the
heterogeneities. Also, with regard to the functional and UDDI specification and enhances its service publication and
processes mediation, they consider it according to the discovery. The authors combine the use of SAWSDL for
syntactic and semantic compatibility of connectivity and creating semantically annotated descriptions of service
don’t present any solution for heterogeneities related to the interfaces and the use of OWL-DL for modelling service
different aspects of the SWS. capabilities and for performing matchmaking via DL
SOPHIE (Arroyo and Sicilia, 2010) is a conceptual reasoning.
framework and architecture for a choreography service In the METEOR-S project (Sheth et al., 2008), four kinds
realised as a SOA. The conceptual model describes the of semantics for web services are defined: data, functional,
structure, behaviour, operation and ontologies as separate non-functional, and execution. These kinds of semantics are
concerns. The behavioural model is defined as Finite State used during the complete lifecycle of semantic web processes.
Machines (FSMs). A set of algorithms are presented to The execution environment handles heterogeneities at protocol
generate mediators for overcoming heterogeneity among the and data level using proxies with data and protocol mediation
MEPs of interacting parties. However, those algorithms are capabilities. For capturing interaction protocols, the UML
not implemented yet and the MEP’s heterogeneities they activity diagrams are called. The data mediation architecture
overcome are not mentioned. (Nagarajan et al., 2006) uses the extensibility features of
Peer-to-Peer/Multi-Agent Systems (P2P/MAS) (Gharzouli WSDL and the popular SOAP engine, Axis2. METEOR-S
and Derdour, 2014) is a generic architecture for the SWS focuses mainly on composition and data mediation.
discovery and composition. The authors show how P2P Bouras et al. (2010) proposed a semantically-enriched
networks can implement open-MAS architectures to build a approach for dynamic data mediation in Enterprise Application
collaborative distributed system. For semantic description of Integration (EAI) scenarios, focusing on the resolution of
web services they use OWL-S and they present behaviours message level heterogeneities between collaborative enterprise
214 K. Mecheri et al.
services, with facilitating the automated data mediation during been defined with the aim to semantically describe web
execution by providing formal transformations of the output services. Authors propose a new approach, Semantic
and input messages to a common reference business data Annotation for MaaS (SA4MaaS) services, as an extension of
model, that is, the ENterprise Interoperability Ontology the W3C recommendation on semantics for web services
(ENIO). (SAWSDL). They use two ontologies: a domain ontology,
Dengping et al. (2011) proposed SAWSDL-iMatcher, a in which the concepts define a semantics of the business domain,
customised and efficient matchmaker of SWS based on and a multimedia ontology, in which the concepts define a set
iXQuery, which extends XQuery with similarity joins. The of annotation properties for describing the multimedia content.
authors showed how iXQuery combines structured queries
with similarity joins to perform a matchmaking SAWSDL 3.3 Discussion
service and how the users can easily customise their preferred
matching strategies in SAWSDL-iMatcher in order to discover In Table 1, we present a classification and a comparison of
the right web services. The ontologies describing the some semantic interoperability works using web services based
functionalities of these services are expressed with OWL. on criteria we have defined. We begin by differentiating them
Midouni et al. (2016) defined a full service approach, according to the semantic description and languages, the
called SA4MaaS, in order to compose Mobility as a service integration approaches, the functions taken into account
(MaaS) services for multimedia data researching. This (publication, discovery, invocation, etc.), the mediation level
approach is based on a four-phase process: description, (data, functions, processes) and to the composition method
filtering, clustering and restitution. Several approaches have (choreography / orchestration) (Sheng et al., 2014).
We distinguish two principal approaches, those based on the benefits of web services, such as flexibility and weak
semantic description language derived from the semantic coupling, unlike the approaches based on the annotations of
web (first seven works in Table 1), and those based on the web service descriptions, that present more advantages. But,
annotations of the web service descriptions, exploiting the contrary to our approach the last six works (as shown in
extensibility elements of the existing languages (last seven Table 1) did not deal with the mediation of processes.
works in Table 1). These annotations present more advantages Since we adopt the context oriented approach, our
than the description languages since they exploit the web conceptual model allows us to have a dynamic semantic
services existing standards. Besides this, they are relatively mediation because it offers the following advantages:
simple and bring more flexibility. Concerning the web
There is not a unique information representation
service integration, we notice that the works that combine
(common ontology). So, the service providers are not
service oriented approaches and extended description-based
required to adapt their local semantics.
approaches are more interesting than the others since they
benefit from advantages of the web services using existing The semantic heterogeneities are explicitly represented.
languages and standards. Then, they can be automatically interpreted and resolved.
Let us notice that, except the works of Mrissa (2007), all
The reference model is linked to the description of the
the others assume the local semantics adaptation of the web
web service concerned by the SAWSDL.
services to the semantics of the shared ontology, to be able
to perform the mediation. Our approach is service oriented and is based on extended
In our previous work (Mecheri et al., 2012), we have descriptions. So, it is flexible, which promotes reuse, thus
given more details on the data mediation based on the preserving the legacy systems thanks to its weak coupling.
notion of context and SAWSDL.
Concerning our present work, we use the method of
Abstract State Machines (ASMs) like works of Domingue et al. 4 Description of the semantic model
(2008), Cimpian and Mocan (2005), Cabral and Domingue
(2005) for the representation and the mediation of the business In Section 3, we have seen that to describe web services,
processes. Cimpian and Mocan (2005) described the annotation-based approaches are the best. For this purpose,
behavioural aspect with the help of non-functional properties. we will propose a semantic model that serves as a reference
However, in this paper, we consider it using the notion of for the annotations of the WSDL documents of the web
behavioural context we have defined. This gives our model services in interaction. This annotation will be supported by
power of expression and flexibility. SAWSDL. Our model is based on the ASM method and the
In summary, the semantic description languages of web notion of behavioural context that we defined.
services are derived from the semantic web. They attempt to So, the role of this model is to capture the semantics
replace existing web services languages to elevate them to of interacting web services and, therefore, it explicitly
SWS. As a result, the approaches based on semantic describes the heterogeneities at the origin of conflicts. This
description languages (seven first works in Table 1) lose the allows automation of mediation to reconcile heterogeneous
216 K. Mecheri et al.
web services. One of the best methods to capture semantics The data semantic model has been described in a
consists in the use of ontologies. previous work (Mecheri et al., 2012). In the following, we
The semantic model is a very important part in our will describe the behavioural semantic model. The others
architecture because it has four fold-focuses: sub-models are out of the scope of this article.
To resolve semantic heterogeneities of data exchanged
between two WS by providing a reference model of 4.1 Behavioural Semantic Sub Model (BSSM)
data semantics. To capture the behavioural properties semantically, we use the
To facilitate discovery process of WS through the ASM method to take advantages of its descriptive power and
formal representation of the functionalities and the non- it’s modelling based rules (Börger and Bernhard, 2008).
functional properties offered by the WS. Therefore, our BSSM is designed according to the ASM
principles. First, it is state based, with the state being the
To resolve heterogeneities of message exchange signature or vocabulary. Second, the change of state is
sequences through a formal behavioural model. modelled by the transition rules that change the values of the
To resolve heterogeneities of non-functional properties functions and relations defined by the signature. The state is
of WS requester and WS provider. described by an ontology exploiting the notion of semantic
object and context. The functions and relations of the ASM are
All those objectives of our semantic model aim to support represented in our model by the notion of semantic object. For
semi-automated process of mediation and consequently the this, we adapt the definition of data context to define the notion
interoperability of WS at the semantic level. To have a full of behavioural context describing the category of the semantic
semantic description of web services, we must consider object.
their functional and non-functional properties as well as
Definition 1: The behavioural context includes the information
their behavioural ones. Our conceptual architecture is built
related to the interaction in which the web service is involved.
according to the four levels of our mediation classification
This information is necessary to the interpretation of the
(see Sub-section 2.3). It is inspired from a semantic model
communication actions.
and is composed of four sub-models (see Figure 2):
Definition 2: A semantic object S is defined as a 5-tuple:
Functional properties semantic sub-model (FSSM):
describes the semantics of the functional properties as S = (c, v, t, DC, BC)
functionalities offered by the web service, the protocols where:
and the encoding (RPC, encoded, Document, and Literal).
c: is a concept of S that defines the ontologic family to
Non-functional properties semantic sub-model (NSSM): which data belongs, that is the abstract class of which the
specifies the semantics of the non-functional properties data is an instance,
of the web services as the properties concerning the
v: is a value of S, that is the data itself: it is an instance of
categorisation, the QoS, the temporal availability, and
the type of the data,
so on.
t: is a type that defines the kind of content of the data
Behavioural semantic sub-model (BSSM): defines the (simple / complex XML schema),
role semantics of the functionality of web service to be
achieved; in our case, it is handled by the behavioural DC: is a data context of S that brings some precisions about
ontologies. the interpretation of the data. It is a link to a data contextual
ontology.
Data semantic sub-model (DSSM): describes the semantics
BC: is a behavioural context of S that brings some
of the data exchanged between two web services.
precisions about the interpretation of the behavioural aspect.
These sub-models will be used respectively in the mediation of: It is a link to a behaviour contextual ontology containing,
behavioural, data, functional and non-functional properties. among other concepts, the following:
Figure 2 Semantic sub-models
Pattern P: describing the category of the semantic existing architectures and the adherence of new providers. The
object (data exchanged at semantic level). According to defined ontologies are expressed in OWL (Schreiber and Dean,
the classification of ASM functions, the behavioural 2004).
context P (Pattern) can have one of the five following The transition rules express the changes of states by the
values: changes of the set of the instances. Each rule is expressed
by conditions and effects that are Description Logic (DL)
o Static: the semantic object S cannot be changed expressions. To express these rules, we use the Service Web
during the execution. Rule Language (SWRL) (Horrocks et al., 2004).
o In (monitored): S cannot be changed by the
environment, and can be read by the machine. 4.2 Mechanisms of semantic annotation
o Out: S cannot be changed by the execution of the In our approach, we use the annotation mechanisms of the
transition rules, not read by the machine and can be SAWSDL in order to semantically enrich both data, to achieve
read by the environment. automated data mediation, and message exchange sequences
and to achieve automated business process mediation.
o Shared: S can be changed by the machine and by
In order to annotate semantically data, the SAWSDL
the environment.
attributes (Lifting Schema Mapping and Lowering Schema
o Controlled: S is only updated by the machine; it Mapping) are added to the source web service and target
cannot be read or updated by the environment. web service descriptions (the XML schema.XSD). These
annotations make reference to the mapping definitions of
Order (Ord): its value is a number that gives the order XML data from and to DSSM.
of the message (exchanged data). For the semantic annotation of behavioural aspects (i.e.
Acknowledgement (Ack): whose the value is true if an MEPs), our approach consists in adding the Model Reference
acknowledgement message is required. attributes to the operations level in the WSDL document. This
attribute make reference to the BSSM previously described.
As shown in Figure 3, our BSSM contains two types of The semantic model (the DSSM and the BSSM) is an
ontologies: Behavioural Local Ontologies (BLO) and important part in the mediation of both data and business
Behavioural Contextual Ontologies (BCO). BLOs describe the processes.
local semantics of the exchanged XML data (XML schema) to
the level of the web service provider. BCOs are associated to
the concepts of the local ontology. A contextual ontology 5 Mediation levels for SWS interoperability
permits to describe the context of a local ontology concept.
Thus, to every local ontology concept is associated a contextual To interoperate SWS, we need tasks of dynamic discovery,
ontology that allows one to clarify the behavioural semantic selection, and invocation. For each of these tasks, mediation
heterogeneities and to solve the process conflicts related to the may be required at different levels: at the data level, at the
user’s contexts at runtime. functional level, at the business process level, and at the
non-functional one.
Figure 3 Structure of the behavioural semantic sub-model In our approach, four levels of mediation are provided
by the four types of web services:
WSDM: Web Service of Data Mediation,
Ack WSFM: Web Service of Functional Mediation,
Ord
WSPM: Web Service of Process Mediation,
P
Behavioural Contextual WSNM: Web Service of Non-functional Mediation.
Ontology (BCO)
As shown in Figure 4, mediation is supported by the four
types of web services above by exploiting the ontologies of
the semantic model described in precedent section as
Ack follows:
Behavioural Local P
Ord Non-functional SWS mediation can be performed by
ontology (BLO) Behavioural Contextual the WSNM using the NSSM.
Ontology (BCO)
The adaptation of interfaces, i.e. the mediation of the
functional properties, can be carried out by the WSFM
Every provider describes the semantics of his own web using the FSSM. The WSFM can call the WSNM.
services. This allows a reliability of the annotations of the web
The exchanged data can be interoperated using the
service description and the integration of the existing services
WSDM using the DSSM.
without need to any supplementary service description. The use
of the SAWSDL languages makes this solution compatible Mediation of business processes can be done by WSPM
with the WSDL, allowing, thus, the interoperability with the using BSSM. WSPM can call WSDM.
218 K. Mecheri et al.
Mediation used for discovery and selection Mediation used during execution
No
Resolved conflicts? Failure
Yes
No
Conflicts?
Yes
No
Resolved conflicts? Failure
Yes
BSSMr BSSMp
WS requestor WS provider
WSPM interface
Inference
WSDL reader Mediator Engine
(Reasoner)
BSSM
Knowledge
BSSM WSDM
Base
220 K. Mecheri et al.
The communication action of the WSPM via its interface is add(ci, BSSM2) /* add ci C of BSSM1
a set of messages corresponding to: to BSSM2 according to context information
(ord) */
Receiving a request of the WS requester. endif;
Sending this request to the WS provider. endfor;
3 semantic- syntactic transformations
Receiving an answer of the WS provider.
send(msg, WS2)
Sending this answer to the WS requester. {for all c C of BSSM2 do
msg= lowringSchemaMapping(c);
The WSPM can perform, with using its capacity of mediation, the
send msg to WS2);
transformations of these messages to ensure the communication
between the web services requester and provider. endfor;
The process mediation algorithm is described in Listing 1. }
2’
Semantic Semantic
Description (OWL): Description (OWL):
3 1 3
1
It should be noted that our main objective is to design values. These mappings are supported by the WSDM
and realise these mediation steps by creating the three (Mecheri et al., 2012).
transformation modules, namely:
o Then, for the messages, these mappings are supported
The Semantic-Semantic Transformation (SST) module: by the WSPM precisely by the SST module, with the
responsible for the semantic to semantic transformations mediation algorithm exploiting the ontologies of the
(arrows 2 and 2ʹ in Figure 8) based on the ontologies BSSM its contextual information as well as the
mappings of the BSSMs. predefined transition rules. The WSPM calls the
The Syntactic-Semantic Transformation (SynSem) WSDM for data mappings.
module: responsible for the syntactic to semantic In the following, we describe the SST module. However, it
transformations (arrows 1 and 1' in Figure 8). should be noted that these modules work together to
The Semantic-Syntactic Transformation (SemSynT) perform the semantic mediation of business processes (and
module: responsible for the semantic to syntactic data) as shown by the algorithm (see Listing 1).
transformations (arrows 3 and 3ʹ in Figure 8).
6.5 Structure and role of the SST module
6.4 Conversion and transformation functions
Our approach to designing the SST module is based on
According to the principle of our mediation of the business the notion of mapping BSSM ontologies and the reasoning
processes several types of transformations are defined on these mappings. It can be summarised as follows
namely: (see Figure 9):
Semantic-syntactic transformations: using SAWSDL Use the BSSM ontologies (BLO and BCO) capturing
“liftingSchemaMapping” annotations associating an the semantics of the two interacting heterogeneous web
XML schema type or element with a mapping to the services (at the process level).
semantic model, defining how an instance of the XML
document is transformed to one or more concepts of the Create mappings between these two ontologies:
semantic model. These are transformations of the data o Fusion of the two ontologies to have the SST
conveyed by the messages. For messages, the SAWSDL ontology.
annotations “ModelReference” are used, associating the
WSDL message with its concept in the BSSM. o Creation of the mapping rules between the
concepts of SST ontology.
Syntactic-semantic transformations: they concern
the data, using the SAWSDL annotations o Use the reasoning technique on these mappings:
“loweringSchemaMapping” associating a concept of execution of rules by an inference engine to make
the semantic model with a mapping to a data of the interoperate these two ontologies and consequently
XML document (using the SPARQL followed by the to resolve the semantic conflicts.
XSLT), defining how a semantic concept is transformed
Into an XML data item. It also involves transformations of Figure 9 illustrates the architecture diagram of the SST
the data conveyed by the messages. module that follows from this approach:
It should be noted that the ontologies of the BSSM must
Semantic-semantic transformations: be syntactically and terminologically compatible, but they
o First, for the data in addition to the classical mappings will be heterogeneous at the conceptual (semantic) level.
of ontologies, we use conversion functions of the The used approach is the multi-ontologies approach.
222 K. Mecheri et al.
Figure 9 Architecture of the SST module Algeria). It is organised as departments and antennas
distributed on the East of the national territory.
Transition The activity of the enterprise is essentially characterised by
SST Ontology Rules (swrl) the control, the expertise, the geometric studies and structural
BSSM conception of the roads, railways, airports and so on.
1
BSSM2
The SETA develops multiple relations with outside
Inference
engine organisms as the analysis laboratories, national and
international suppliers, the banks, and so on.
BSSM1 BSSM2 For its work on construction sites SETA buys topographic
(BLO2+BCOs2)
(BLO1+BCOs1) and surveyors equipment’s from international suppliers.
The Information System (IS) of SETA is service-oriented.
We assume that the SETA partners make this transition to
SOA, and they also offer business services (Mecheri et al.,
WSDL1 WSDL2
WS2 2010).
WS1
The business services implemented by the semantic web
services are exposed to the partners of the SETA, on the other
6.5.1 Transition rules hand, these last will benefit from the services offered by its
partners. It is therefore a cooperation of information systems.
In our case, the rules make it possible to instantiate concepts To evaluate how our approach overcomes the
of an ontology with the instances of the concepts of another
heterogeneities between SWS at behavioural level, we use two
ontology, we can have five cases whose first three lead to
web services:
transitions that is to say change of state:
Requester web service (of SETA IS), asking for a quote
A concept (with P = out) of ontology1 instantiates a
concept (with P = in) of ontology2 (identical concepts). concerning some products (topographic equipment) and
send the order.
A concept (with P = out) of ontology1 instantiates two
concepts (with P = in) of ontology2 (split), that is to say Provider web service (of supplier) that answers by a
a concept of ontology1 equivalent to two concepts of quote of the requested products and invoice.
Ontology2. The message exchange sequences between these two web
Two concepts of ontology1 (with P = out) instantiate a services in interaction are described in Figure 10.
concept (with P = in) of ontology2 (merge), i.e. two According to the choreography given in this figure the
concepts of ontology1 equivalent to one concept of behavioural interfaces of the two web services are given in
Ontology2. Table 2.
If a concept of ontology1 with P = out which does not
instantiate any concept of ontology2, that is, this concept 7.2 Implementation and tools
of ontology1 is not equivalent to any concept of For implementing the semantic model (ontologies), we
ontology2, the mediator does nothing for this concept.
chose the following:
A concept among the first three cases of ontology1 has
Protégé 3.4.1 for ontology’s creation and edition
its contextual information for Ack = True, i.e. it needs
an acknowledgment, which will be generated by the (Protégé, 2017), with the
mediator. SWRLTab plugin for SWRL rules edition (Horrocks
et al., 2004),
7 Some implementation aspects and a case study Protégé-OWL API for access to ontologies by program
(Schreiber and Dean, 2004),
7.1 Case study Pellet (2017) as inference engine and
The main objective of our research is to make interoperate Jess (2017) for rules execution .
the information system applications of SETA (Technical
Studies Society of Annaba) between them and with those of These choices are justified by the need to use some custom
its environment. It is about intra-enterprise and inter- SWRL built-ins only available in the Pellet Protégé 3.4 and
enterprises interoperability (Mecheri and Souici-Meslati, 2010; Jess. We use Pellet for the other reasoning tasks (coherence
Mecheri et al., 2010, 2012). verification or classification). The access to Pellet and to
The SETA is an economic and autonomous public Jess from Protégé-OWL API is based available specific
company, whose head office is fixed in Annaba (East of bridges.
Context-based interoperability of semantic web services 223
QuoteRequest
QuoteReqProduct
QuoteId
QuoteResponse
WS WS
QuotePrice
requester provider
WSPM
Order Order
Ack
InvoicePrice InvoiceId
InvoiceId InvoicePrice
7.2.1 Build of behavioural semantic sub models data conflicts (names, types, and so on). These are the
(BSSMs) transitions rules used in our case study.
To build the two BSSMs of WS requester and WS provider, Transition rules of WS requester corresponding to BLO
we use Protégé to create Behavioural Local Ontologies Requester (p1):
(BLOs). For each concept of BLO, we add its Behavioural o Rule 1: the WS requester creates an instance of
Contextual Ontology (BCO) (see Figure 11). ‘quoteRequest’ from the instances for which it has
‘quoteReqId’ and ‘quoteReqProduct’. This instance
7.2.2 Creation of SST ontology and mapping rules of ‘quoteRequest’ is sent to WSPM:
To perform the mediation between the ontologies (BLO quoteReqId (?QRId) quoteReqProduct (?QRP)
Requester and BLO Provider) of the two Behaviourist quoteRequest(?x[?QRId, ?QRP])
Models (BSSMs), we create the ‘SST Ontology’ by merging
the first two. Here, we used OWL: import function to import Transition rules of WS provider corresponding to BLO
the two ontologies: ‘BLO Requester’ and ‘BLO Provider’ in Provider (p2):
‘SST Ontology’. This allows us to manipulate both at the
o Rule 2, Rule 3: The WS provider creates a ‘quoteId’
same time; using SWRL mapping rules that we created and
then successfully executed. instance and a ‘quotePrice’ instance after
We have integrated the JESS rule engine with the SWRL having received a ‘quoteReqId’ instance and a
language editor in Protege to execute these rules. In order to ‘quoteReqProduct’ instance from WSPM. The
keep the presentation simple, we assume that there are no created instances are sent to WSPM:
224 K. Mecheri et al.
quoteReqId (?QRId) quoteReqProduct (?QRP) quoteId p2:QuoteId(?X)^p2: QuotePrice(?Y)^ ?R= swrl :concat
(?x) ( ?X ?Y) → p1 :QuteResponse( ?R).
quoteReqId (?QRId) quoteReqProduct (?QRP) This rule combines the two concepts ‘QuoteId’ and
quotePrice (?x) ‘QuotePrice’ of the BLO Provider prefixed by p2 to send
them to the BLO Requester prefixed by p1 in the concept
As example of an alignment rule, WSPM combines the ‘QuoteResponse’ (see Figure 12). That is, both the
outputs of WS provider and sends it to WS requester, in ‘QuoteId’ and ‘QuotePrice’ concepts of p2 are equivalent to
order to obtain a quote response: a single ‘QuoteResponse’ concept of p1.
Figure 11 BSSMs of WS Requester and WS Provider
Gharzouli, M. and Derdour, M. (2014) ‘To implement an open- Midouni, S.A.D., Amghar, Y. and Chikh, A. (2016) ‘A semantic
MAS architecture for semantic web services discovery, what annotation for MaaS services: the SA4MaaS approach’,
kind of P2P protocol do we need?’, International Journal of International Journal of Metadata, Semantics and Ontologies,
Agent Technologies and Systems (IJATS), Vol. 6, No. 3, Vol. 11, No. 2, pp.82–92.
pp.57–72. Miller, J., Verma, K., Rajasekaran, P., Sheth, A., Aggarwal, R. and
Haller, A., Cimpian, E., Mocan, E., Oren, E. and Bussler, C. Sivashanmugam, K. (2004) WSDL-S : Adding Semantics to
(2005) ‘Wsmx – a semantic service-oriented architecture’, WSDL, White Paper, Technical Report, Large Scale
International Conference on Web Services, Orlando, Florida, Distributed Information Systems. Available online at:
USA. http://lsdis.cs.uga.edu/library/download/wsdl-s.pdf.
Hausberger, T. (2009) ‘Semantic Annotations for WSDL, Semantic Monfort, V. and Goudeau, S. (2004) Web services et interopérabilité
Web Services PS, Slides, Departement of Computer Science des systèmes d’information, Dunod, Paris, France.
Leopold Franzens University Innsbruck. Available online at: Mrissa, M. (2007) Médiation Sémantique Orientée Contexte pour la
http://www.schneier.com/essay-037.pdf Composition de Services Web, Thèse de Doctorat, Université
Horrocks, I., Patl-Schneider, P.F., Boley, H., Tabet, S., Grosofn, B. Claude Bernard Lyon I.
and Dean, M. (2004) SWRL: A Semantic Web Rule Language Nagarajan, M., Verma, K., Sheth, A.P., Miller, J. and Lathem, J.
Combining OWL and RuleML, W3C, Technical Report, (2006) ‘Semantic interoperability of web services – challenges
Available online at: http://www.w3.org/ Submission and experiences’, Proceedings of the IEEE International
/2004/SUBM-SWRL-20040521/ (accessed on April 2017). Conference on Web Services (ICWS’06), Chicago, USA.
Izza, S., Vincent, L. and Burlat, P. (2008) ‘Exploiting semantic Paolucci, M., Kawamura, T., Payne, T. and Sycara, K. (2002)
web services in achieving flexible application integration in ‘Importing the semantic web in UDDI’, International
the microelectronics field’, Computers in Industry, Vol 59, Workshop on Web Services, E-Business, and the Semantic
pp.722–740. Web (WES’02), Toronto, Canada, Vol. 2512, pp.815–821.
JESS (2017) Website. Available online at: http://herzberg. Peer, J. (2005) ‘Semantic service markup with SESMA’. Web
ca.sandia.gov/jess/ (accessed on April 2017). Service Semantics Workshop (WSS'05) at the 14th
International World Wide Web Conference (WWW'05).
Khater, M., Habibeche, S. and Malki, M. (2017) ‘Behaviour
Pellet (2017) Available online at: https://github.com/stardog-
approach for composite OWL-S services discovery’,
union/pellet (accessed on April 2017).
International Journal of Business Information Systems,
Vol. 25, No. 1, pp.55–70. Pengcheng, Z., Huiying, J., Zhipeng, H., Hareton, L, Wei, S. and
Yan, J. (2018) ‘IgS-wBSRM: a time-aware web service QoS
Klein, M., König-Ries, B. and Müssig, M. (2005) ‘What is needed monitoring approach in dynamic environments’, Information
for semantic service descriptions – a proposal for suitable and Software Technology, Vol. 96, pp.14–26.
language constructs’, International Journal on Web and Grid
Services (IJWGS), Vol. 1, Nos. 3/4, pp.328–364. Protégé (2017) Website. Available online at: http://protege.stanford.
edu/ (accessed on April 2017).
Kourtesis, D. and Paraskakis, I. (2008) ‘Combining SAWSDL,
Schreiber, G. and Dean, M. (2004) OWL web ontology language
OWL-DL and UDDI for semantically enhanced web service
reference. Available online at: http://www.w3.org/TR/2004/
discovery’, in Bechhofer, S., Hauswirth, M., Hoffmann, J.,
REC-owl-ref-20040210/ (accessed on April 2017).
Koubarakis, M. (eds): ESWC, LNCS 5021, Berlin/Heidelberg,
Germany, pp.614–628. Sheng, Q.Z., Qiao, X., Vasilakos, A.V., Szabo, C., Bourne, S. and Xu,
X. (2014) ‘Web services composition: a decade’s overview’,
Martin, D., Burstein, M., McDermott, D., McIlraith, S., Paolucci, Information Sciences, Vol. 280, pp.218–238.
M., Sycara, K., McGuinness, D.L., Sirin, E.N. and Srinivasan,
Sheth, P.A., Gomadam, K. and Ranabahu, A. (2008) ‘Semantics
N. (2004) ‘Bringing semantics to web services with OWL-S’,
enhanced services: METEOR-S, SAWSDL and SA-REST’,
World Wide Web, Vol. 10, pp.243–277.
Data Engineering Bulletin Issues, IEEE Technical Committee,
Mecheri, K. and Souici-Meslati, L. (2010) ‘Semantic Vol. 31, pp.8–12.
interoperability of web services: a survey’, Proceedings of the
Shridevi, S. and Raju, G. (2018) ‘A literature survey on the
1st IEEE International Conference on Machine and Web
performance evaluation model of semantics enabled web
Intelligence (ICMWI’10), Algiers, Algeria, pp.55–60.
services’, International Journal of Advanced Intelligence
Mecheri, K., Boufaida, M. and Souici-Meslati, L. (2010) ‘Une Paradigms, Vol. 10, Nos. 1/2, pp.160–177.
architecture orientée services pour l’interopérabilité des SI
dans la SETA’, Proceedings of the Second Workshop on Web Sikos, L.F. (2015) Mastering Structured Data on the Semantic Web,
Services (WWS’10), CERIST, Alger, Algérie, pp.37–48. 1st edition, Apress.
Mecheri, K., Boufaida, M. and Souici-Meslati, L. (2012) ‘Data Taher, Y., Benslimane, D., Fauvet, M-C. and Maamar, Z. (2006)
mediation towards semantic web service interoperability’, ‘Towards an approach for web services substitution’, in Ghodous,
International Journal of Web Science, Vol. 1, No. 3, pp.163–176. P., Dieng-Kuntz, R., Loureiro, G. (eds): IDEAS, pp.166–173.
Medjahed, B., Benatallah, B., Bouguettaya, A. and Elmagarmid, Tausan, N., Markkula, J., Kuvaja, P. and Oivo, M. (2017)
A.K. (2004) ‘Webbis: an infrastructure for agile integration of ‘Choreography in the embedded systems domain: a systematic
web services’, International Journal of Cooperative literature review’, Information and Software Technology, Vol. 91,
Information Systems, Vol. 13, No. 2, pp.121–158. pp.82–101.
Metrouh, A. and Mokhati, F. (2018) ‘Flexible web services Vitvar, T., Zaremba, M., Mocan, M. and Mocan, A. (2009) ‘Mediation
integration: a novel personalised social approach’, Journal of using WSMO, WSML and WSMX’, in Petrie, C. et al. (eds):
Experimental and Theoretical Artificial Intelligence, Vol. 30, Semantic Web Services Challenge Results from the first year,
No. 3, pp.441–456. pp.31–49.