AN APPROACH TO DATA-DRIVEN ADAPTABLE SERVICE
PROCESSES
George Athanasopoulos and Aphrodite Tsalgatidou
Dept. of Informatics & Telecommunication, National and Kapodistrian University of Athens, Athens, Greece
Keywords: Adaptable Systems, Service Compositions, Data-driven Adaptations, Context-aware Software Systems.
Abstract: Within the currently forming pervasive computing environment, services and information sources thrive.
Instantiations of the service oriented computing paradigm, e.g. Web, Peer-to-Peer (P2P) and Grid services,
are continuously emerging, whilst information can be collected from several information sources, e.g.
materializations of the Web 2.0 and Web 3.0 trends, Social Networking apps and Sensor Networks. Within
this context the development of adaptable service oriented processes utilizing heterogeneous services, in
addition to available information, is an emerging trend. This paper presents an approach and an enabling
architecture that leverage the provision of data-driven, adaptable, heterogeneous service processes. Core
within the proposed architecture is a set of interacting components that accommodate the acquisition of
information, the execution of service chains and their adaptation, based on collected information.
1 INTRODUCTION
Services and information sources flourish within the
currently forming pervasive computing environment.
Service-Oriented Computing (SOC) along with the
emerging instantiations, e.g. Web, P2P and Grid
services, promise to revolutionize the way
applications and systems are built by fostering the
provision of adaptable systems. In addition, the
emerging Sensor Web (Botts et al., 2008), and the
materializations of the Web 2.0 and Web 3.0
paradigms, e.g. Social Networking applications or
Agent based systems, provide new types of
information sources that can be exploited towards
the provision of emerging types of systems and
services. Within this frame, the development of
adaptable service processes, comprising
heterogeneous services and information stemming
from existing sources, is an emerging trend; in the
context of this paper, service processes are regarded
as systems, which operate within a specific
environment, and perform pre-specified activities,
via the use of services, in an orderly manner
producing and/or consuming related information
during their execution. This trend, cannot be
addressed by contemporary rigid approaches such as
WS-BPEL (Alves, et al., 2007). Therefore, the
development of adaptable service processes calls for
novel approaches.
AI techniques (especially AI planning ones) have
been extensively applied towards the provision of
dynamically composed service oriented
orchestrations (Jinghai & Xiaomeng, 2004). Most of
them focus on the provision of automatically
constructed orchestrations (or similarly called task
plans) that comprise Web services solely. Along the
same lines, the Context Aware Computing (CAC)
research community has applied considerable efforts
towards the utilization of contextual information for
the adaptation of web service compositions, e.g.
(Lirong, Zhongzhi & Fen, 2006). The majority of
these approaches in both research fields address
neither the interoperability concerns raised by the
multiple instantiations of the service oriented
computing paradigm nor the utilization of available
and/or emerging information sources in tandem.
An emerging approach towards the integration of
services that has lately received considerable
momentum is based on the utilization of the
Tuplespace model (Rossi, Cabri & Denti, 2001).
Within this paper we present an approach which
leverages the merits of tuplespace paradigm for the
provision of data-driven, adaptable, heterogeneous,
service compositions. A component, implementing
the Tuplespace model, called Semantic Context
Space Engine (SCS Engine), along with a Service
Orchestration Engine and appropriate adaptation
139
Athanasopoulos G. and Tsalgatidou A. (2010).
AN APPROACH TO DATA-DRIVEN ADAPTABLE SERVICE PROCESSES.
In Proceedings of the 5th International Conference on Software and Data Technologies, pages 139-145
DOI: 10.5220/0003007401390145
Copyright
c
SciTePress
algorithms (i.e. Process Optimizer) provide the
required functionality.
The rest of the paper is organized as follows:
next we present an approach towards the provision
of such a mechanism and the prime components of
the proposed architecture; in the following we
present a comparison of our approach against similar
approaches. We conclude this paper with the
presentation of our remarks on the presented
approach and our future work plans.
2 DATA-DRIVEN ADAPTATION
The prime assumption of our approach is based on
the observation that a service process, comprising
heterogeneous services, should be able to utilize the
information available within its environment and
adapt its execution accordingly. Within this context,
a process state is not solely depending on the values
of its internal parameters, on the resulting outcomes
from the invocation of its constituent services and/or
on its internal operations but, also on information
pertaining to the process environment; therefore, the
latter should be taken into account during process
development and execution.
To facilitate the provision of such adaptable
processes we propose an approach that leverages
information contained within a specific ‘space’ to
adapt a service process using appropriate adaptation
algorithms. A space is considered to be the process’s
environment which is open to other processes and
systems e.g. Service-Oriented systems or Agent
based systems, for information exchange. For the
process to be able to exploit the available
information, appropriate extensions are pre-
configured and embedded within the process
specification.
illustrates the proposed platform with the
comprising components which are:
A Semantic Context Space Engine responsible
for the collection of contextual information,
A Process Optimizer responsible for the
adaptation of service processes based on the
collected information, and
A Service Orchestration Engine responsible for
the execution of heterogeneous service processes.
The SCS Engine provides an open space where one
may place relevant information. From a functional
point of view the SCS Engine leverages one to i)
Write and Retrieve information within the process’s
environment and to ii) Logically Group information
of interest to a specific domain and Specify
associations among logical information groups,
which contain information elements from
related/depending domains.
The Process Optimizer component implements
an AI planner that facilitates the discovery of
process plans that control the execution and
adaptation of service processes. Such plans are
usually modelled as conditional plans which contain
branching control structures i.e. if-then-else that
decide which execution path will be followed based
on the value of a condition. The problem of data-
driven adaptation can be modelled as a non-
deterministic, partial observability planning
problem, where Model Checking techniques (Nau,
Ghallab & Traverso, 2004) are extensively applied
lately.
The Service Orchestration engine provides a
BPEL-based engine that facilitates the execution of
heterogeneous service orchestrations (e.g. Web, Grid
and P2P service orchestrations). One of the core
features of the orchestration engine is the support for
the monitoring and reconfiguration of process state
according to the suggestions made by the Process
Optimizer. To facilitate the execution and the
adaptation of a process state, the orchestration
engine exchanges information with the SCS Engine.
Details on the comprising components are
presented next.
3 SEMANTIC CONTEXT SPACE
The Tuplespace model has been extensively applied
in the coordination of distributed and parallel
systems (Rossi, Cabri & Denti, 2001). Yet, its
utilization by the Service-Oriented Computing
paradigm is an emerging trend that has been
accompanied by proprietary extensions (Nixon et al.,
2008) (Zeng, Lei & Chandramouli, 2005). Although
services provide a layer of abstraction that facilitates
the interoperation of systems over the web, the
merits of the Tuplespace paradigm can further
enhance the Service Oriented model. These merits
include the i) decoupling of process components, the
ii) associative based addressing (i.e. data is
referenced by its content and not by its address) and
the support for the provision of iii) synchronous and
asynchronous communication patterns (Rossi, Cabri
& Denti 2001). More to that, such properties will
foster the formation of new types of collaboration
schemes among Service-Oriented systems and other
existing or emerging types of systems such as Agent
based systems, Sensor and Grid applications.
ICSOFT 2010 - 5th International Conference on Software and Data Technologies
140
Figure 1: Platform high level architecture.
The SCS Engine incorporates appropriate semantic
annotations to the Tuplespace information model
and has been structured so as to accommodate the
following functional needs:
- Support for multiple meta-information models
along with associated querying mechanisms for
the discovery of contained information elements
(i.e. registered pieces of information within the
SCS Engine). Such models could range from
OWL based annotations to simple key-value
pairs.
- Use of ‘scope’ delimiters to facilitate the logical
organization of registered information elements
in ‘information islands’ with related
information.
- Support for the specification of affiliations
among scopes that may spawn across dispersed
engines, facilitating the formation of logical
associations among related ‘information
islands.’
- Use of a Web-based reference mechanism for
uniquely addressing elements and scopes.
- Handling of both Java and XML based
information elements.
In order to facilitate the abovementioned features we
decided to use the JavaSpace service (Freeman,
Hupfer & Arnold, 1999) of the Jini framework
(Waldo & Team, 2000) as a basis for the
implementation of the SCS Engine. The selection of
the JavaSpace service instead of an open source
Tuplespace e.g. TSpace (Fontoura et al., 2003) or an
XML based Tuplespace such as XMLSpace
(Tolksdorf, Liebsch & Nguyen, 2004), was primarily
driven by our need for flexibility over the supported
meta-information models and the related query
engines, as well as the need for supporting Java and
XML based information elements. Nonetheless, one
may argue that the SCS Engine could be
implemented on top of a database management
system e.g. an RDBMS such as Oracle, or MySQL
etc. but the rigidness of their information models and
their complex implementation does not render them
a first class choice.
The interface of the SCS Engine is an extended
version of the Linda model catering for: the writing,
reading (discovery) and retrieval (taking) of
information elements, either from the whole space or
from a specific scope; the persistent querying
(continuous querying) for information elements; and
the management of scopes such as the creation,
discovery, and removal of a scope, the creation
and/or removal of affiliations among scopes. As one
may easily note the provided operations
accommodate a basic set of functionality. This was
decided so as facilitate the extensibility of the SCS
engine upon which additional higher level
operations and mechanisms can be built.
All these operations are accessible either through
a Java based interface or via the use of a Web
service based interface (currently only the Java
based interface has been implemented).
4 PROCESS OPTIMIZATION
As it has been pointed out in other research efforts
too (Pistore et al., 2004), automated service
composition can be mapped to a non-deterministic,
partially observable planning problem. This is
because, similar to a partially observable planning
problem, automated service composition has to
confront concerns such as the uncertainty raised by
the interaction with external services (and partners)
and the partial knowledge of the composition state
accruing from the lack of information on the
internals of each constituent service (i.e. interacting
partner). The construction of task plans in non-
deterministic and partially observable domains has
received considerable investigation by the AI
planning community, e.g. (Kuter et al., 2007)
(Bryce, 2006). Solutions to such problems are in the
form of conditional plans; conditional plans contain
branching control structures i.e. if-then-else
structures that decide which path will be followed
based on the value of a condition.
In this frame, a formal representation of the
problem domain (D) is a tuple
,,, ,
, where:
AN APPROACH TO DATA-DRIVEN ADAPTABLE SERVICE PROCESSES
141
- S: is the finite set of states of the associated state
transition system,
- Α: is the finite set of actions 
|
,
-  is the transition relation,
- : is a finite state of observation variables

| in
-
:
,
is the relation for the evaluation of
observation variables  on each state.
Within our context the value of an observation
variable is independent of the action that may
have preceded
Contrary to what stands in a deterministic problem
domain (Nau, Ghallab & Traverso, 2004) the
transition relation R can map the execution of an
action , on a state  (assuming that α is
applicable on s) to more than one successor states
i.e.
,
1. An action α is applicable on a
state  iff there exists a state  such that
,,
stands. The set O contains the finite set of
observation variables o
i
whose values are evaluated
at runtime. The value of each observation variable at
each state is defined by the observation relation Χ. A
simplification normally introduced to avoid the
unnecessary complexities is to consider observation
variables as boolean variables whose values could be
either true or false (i.e.
,
. Therefore if
,
holds at a state  then the value of variable o at
state s is True. The dual holds in cases where
variable o is False. In cases where both
,
and
,
hold variable o has an undefined value.
Even though the data-driven process adaptation
problem can be mapped to the non-deterministic,
partially observable planning problem a thorough
look into the requirements of the proposed approach
unveils additional concerns that are not properly
handled by the classic representation of the planning
problem (D) and the associated solutions e.g.
(Pistore et al., 2004)(Pistore et al., 2005). These
concerns are related to the ‘origin’ and ‘validity-
time’ property of observed information, as well as to
the need to consider additional information (i.e.
partially matching information) to the one that is
captured by the pre-specified set of observation
variables.
Most of the existing approaches, adhering to D,
utilize observation variables to grasp information
stemming from either the controlling process or
partner services. Nonetheless, this assumption
hinders the interaction of the controlling process
with external (i.e. with respect to the process)
systems and information sources. Observation
variables are tuned for collecting information from
pre-specified sources i.e. constituent services or the
controlling process, thus neglecting information that
may stem from uncontrolled sources. Within a
controlled environment observation variables are
always conforming to specific constraints expressed
in terms of format and semantic meaning but, in a
pervasive computing environment varying format
and semantic meaning is the norm rather than the
exception. An additional important information trait
is that of the associated validity-time; validity-time
dictates the period of time within which information
may be safely consumed. Existing approaches
adhering to D, avoid considering this property when
dealing with the automated composition of services.
To cater for these concerns we propose a set of
appropriate extensions and supporting mechanisms.
These comprise: a) a mechanism facilitating the
valuation of observations (i.e. observation variables)
based on queries executed over an open set of
information elements, b) an observation
‘interpolation’ mechanism that facilitates the
inclusion of additional information elements, and c)
the use of validity-time property for each
information element and appropriate management
features.
Details on the proposed extensions and their
ramifications on the planning domain are presented
next.
4.1 Observation Considerations
We can safely assume that an observation variable
(o
n
) of the planning domain D is defined in a finite
set of information elements i.e.

(see
(Pistore et al., 2004)). The assessment of an
observation variable at runtime should always return
a value within the specified set OD
n
. In the frame of
our approach the valuation of an observation
variable (o
n
) is mapped to the assessment of a query
(q
n
) performed over an information source, which
corresponds to the system environment. However,
due to its open nature, such an information source
may comprise information elements that are
irrelevant to an executing process. Thus, the set of
queries (Q={q
1
,q
2
,..,q
n
}) performed over a source
should be properly structured so as to avoid the
retrieval of erroneous information elements. To
accommodate this concern we provide a semantic-
based information discovery mechanism. Queries
(q
i
) are extracted out of the semantic and syntactic
details of the associated observation variables (o
i
).
The SCS Engine, executes these queries, and
matching results are returned back to the running
process.
Our approach towards the use of similar
information elements is based on the interpolation of
ICSOFT 2010 - 5th International Conference on Software and Data Technologies
142
observations with related ones in a manner
controlled by the utilized ontology. The
underpinning assumption is that instead of
considering partially matching results to the
performed observations we may as well look for
exact matches to ‘partially matching’ observations.
Hence, the set of observations Ο linked to a process
is expanded by the introduction of related
observations. To facilitate this process we introduce
two additional features i.e. an expansion operator
(oExp) and an expansion ratio property (sD). sD
dictates the minimum (i.e. the infimum) similarity
distance among the concepts of an ontology so as to
consider them part of the same set (i.e. expansion
set). Given an ontology, a concept (co) of that
ontology and an expansion ration (sD) the oExp
operator returns all terms of the provided ontology
whose similarity to co is equal to or greater than sD.
The original set of observations Ο associated to a
specific process can therefore be expanded to a set
Ο΄ with the application of the oExp operator over O
for a specific sD value and ontology Vc. To ensure
that the extended set (Ο' ) contains observations that
can be handled by our system we need to prune it so
as to remove observations that may lead to
unacceptable states. The pruning process ensures
that Ο' contains observations leading to states where
actions (α) belonging to A can be applied.
Finally, to accommodate the concern related to
the information validity time, every information
element registered to the SCS Engine is attributed
with a specific Time To Live (TTL) property which
dictates its validity period. The SCS Engine is
responsible for cleaning up the space and removing
obsolete information elements. Therefore, executing
queries for the discovery of information are ensured
to return related and valid information elements.
5 PROCESS EXECUTION
The outcome of the process optimization procedure
is a conditional plan, containing possible adaptation
cases, that can be transformed into an enhanced WS-
BPEL description. The pre-defined adaptation cases
incorporated with the enhanced service specification
control the adjustment of a process based on
collected information. The provided specification
can be seamlessly executed by an existing BPEL
based orchestration engine.
Nonetheless, the need to accommodate the use of
several types of services poses significant concerns.
This requirement has been exemplified in the
SODIUM project (Tsalgatidou, et al., 2008); our
approach towards the accommodation of
heterogeneous services has been influenced by the
outcomes of this project. Appropriate extensions
have to be introduced both to the Service
Orchestration Engine and to the BPEL specification
to accommodate the invocation of services such as
Grid and P2P services. Nevertheless, as these types
of services i.e. P2P and Grid services, are described
in WSDL-based languages i.e. WSRF (Czajkowski,
et al., 2004) specification and PSDL (Tsalgatidou, et
al., 2008), the required extensions can be easily
accommodated.
PSDL along with the associated middleware
facilitating the description and invocation of P2P
services respectively, can be seamlessly utilized by
existing BPEL orchestration engines to leverage the
use of P2P services. Appropriate plug-ins can be
provided to support the binding of such services. A
crucial aspect for the utilization of Grid services (or
similarly WSRF services) is the identification of
WS-Resources (i.e. the combination of Web
Services and Resources). An indirect approach such
as the one employed in (Ezenweoye et al. 2007),
catering for the specification of the correct address,
i.e. based on the WS-Addressing protocol, can be
used. The WS-Resource endpoint can be provided
either via a variable or as the outcome of a WSRF
Factory service.
Another core feature of the adopted orchestration
engine, rooted to the need for data-driven adaptation,
is the inherent support for the continuous monitoring
of executing process instances. Upon the discovery
of related information by the SCS Engine a process
monitoring component adapts the execution path
according to the suggestions embedded in the
process specification by the Process Optimizer.
Interactions among the Service Orchestration
Engine and the SCS Engine components are
bidirectional. The Service Orchestration Engine
provides information to the SCS Engine that is used
for refining the specified information retrieval
queries, whereas the SCS Engine upon the discovery
of matching information elements pushes them to
the Service Orchestration Engine. Based on certain
criteria, e.g. if the discovered information can be
used for a meaningful adaptation of the running
process, the Process Monitor component of the
Service Orchestration Engine utilizes this
information to adapt t he executing process
accordingly.
An engine that can easily serve as the basis for
the provision of such features is the Apache ODE
engine. Apache ODE is an open source, Java-based
implementation of a BPEL-based orchestration
AN APPROACH TO DATA-DRIVEN ADAPTABLE SERVICE PROCESSES
143
engine facilitating the execution of WS-BPEL v2.0
specifications. It accommodates appropriate
extensions catering for the interaction of the runtime
with external data sources e.g. databases. This
feature can be further extended in order to facilitate
the interactions of the Orchestration Engine and the
SCS Engine.
6 RELATED WORK
The work presented in this paper lies across the
fields of several research communities. The Context
Aware Computing community has spent
considerable efforts on adaptable service
compositions. Contrary to our approach, which
facilitates heterogeneous service processes, quite a
lot of efforts such as the one by Lirong et al.
(Lirong, Zhongzhi & Fen, 2006) have focused on the
provision of adaptable Web service compositions.
Other approaches such as the one employed by
Vukovic, et al. (Vukovic & Robinson, 2004) utilize
contextual information for the elicitation of specific
properties e.g. location, network connectivity, etc.
and consequently adapt web service compositions in
a (semi-) automatic manner.
AI and workflow based techniques have been
also applied for the provision of adaptable service
compositions. Most of these techniques as it has
been also pointed out by Jinghai et al. (Jinghai &
Xiaomeng, 2004), have focused on processes which
comprise Web services or semantically-enhanced
Web services. Further to that, most of these
approaches neglect the importance of contextual
information. Service compositions are usually
modeled as deterministic state transition systems and
the resulting composition problems are modeled as
planning problems within deterministic and fully
observable domains (Nau, Ghallab & Traverso,
2004).
Similar to the approach ensued by Pistore et.al
(Pistore et al., 2005) which caters for the provision
of automated compositions of web services, our
approach considers the provision of composite
services as a non-deterministic, partially observable
planning problem. Nonetheless, we provide for the
use of additional types of services and the
management of contextual information e.g. the
consideration of partially relevant information and
the use of the TTL property, as well as for the use of
other types of service beyond Web services. Along
the lines of the ALLOW project (Marconi et al.,
2009.) our mechanism is also based on the
incorporation of context adaptation mechanisms in
the process specification. Nonetheless, ALLOW
accommodates a constrain based mechanism to
facilitate the evaluation of design-time specific
contextual properties, whereas our approach adheres
to an automatically configured data-based adaptation
strategy.
With respect to existing approaches catering for
the provision of semantically enhanced Tuplespaces,
the SCS Engine provides a flexible and extendable
architecture. Most of the contemporary semantic
Tuplespaces can be considered as Knowledge Bases,
which utilize RDF for the representation of semantic
information. Contrary to these efforts, the SCS
Engine provides a simpler, yet extensible model,
which can support various meta-information
schemes for the annotation of Information elements.
Additional high level functions e.g. automated
inference mechanisms can be built on top of the
proposed mechanism. Further to that, the SCS
Engine accommodates an enhanced logical grouping
mechanism i.e. scopes, which differs considerably
from the ones employed by other contemporary
approaches. Contrary to the scoping mechanism
used by Merrick (Merrick & Wood, 2000) or the one
used in the Semantic Web Spaces (Tolksdorf, et al.,
2005), the provided mechanism accommodates a
flexible affiliation scheme that facilitates the
specification of various types of relationships among
scopes e.g. hierarchical relationships or other user
defined associations.
7 CONCLUSIONS
The need for the provision of data-driven adaptable
service processes has been highlighted in several
application domains, e.g. the crisis management or
the environmental services domain (e.g. Envision
EU project www.envision-project.eu). In this paper
we presented an approach and an enabling
architecture addressing this problem. Core within the
specified approach is a set of components, which
facilitate the collection of information elements, the
adaptation of service processes and their respective
execution.
The proposed approach along with the related
platform constitutes a departure from the current
state of the art. On the one hand it accommodates the
provision of data-driven adaptable heterogeneous
service processes, something that, to the best of our
knowledge, has not been addressed so far whilst, on
the other hand it fosters the opening of service
processes to external systems and the emergence of
additional collaboration schemes among them. This
form of cooperation further promotes the decoupling
among collaborating parties which is an important
ICSOFT 2010 - 5th International Conference on Software and Data Technologies
144
prerequisite for the provision of adaptable systems.
Our future work plans include the finalization of the
implementation of the platform components that are
currently in an early development stage i.e. the
Service Orchestration Engine and the Process
Optimizer. In addition, with respect to the Process
Optimizer, we plan to utilize and evaluate
contemporary algorithms proposed for the respective
planning problem domain e.g. MBP (Pistore et al.,
2004), POND (Bryce, 2006).
ACKNOWLEDGEMENTS
This work has been partially funded by ELKE
(contract 70/3/10294) and the European Commission
(contract ICT-FP7-249120 ENVISION project)
REFERENCES
Alves, A., et al. 2007, Web Services Business Process
Execution Language Version 2.0. OASIS Standard,
Apr. 2007, OASIS WSBPEL TC.
Botts, M., et al., 2008, 'OGC Sensor Web Enablement:
Overview and High Level Architecture', White Paper,
Open Geospatial Consortium INC, (OGC 07-165).
Bryce, D. 2006, 'POND: The Partially-Observable and
Non-Deterministic Planner', International Conference
on Automated Planning & Scheduling 2006, Jun.
2006, Cumbria, UK.
Czajkowski, K., et al., 2004, 'The WS-Resource
Framework', Whitepaper, Globus Alliance.
Ezenweoye, O., et al. 2007, 'Grid service composition in
BPEL for scientific applications', International
Conference on Grid computing, high-performAnce
and Distributed Applications, 2007, Vilamoura,
Algarve, Portugal.
Fontoura, M., et al. 2003, 'TSpaces services suite:
Automating the development and management of Web
Services', 12th International World Wide Web
Conference, 2003, Budapest, HUNGARY.
Freeman, E, Hupfer, S & Arnold, K 1999, JavaSpaces
Principles, Patterns, and Practice: Principles, Patterns,
and Practice, Addison-Wesley.
Jinghai, R. & Xiaomeng, S. 2004, 'A Survey of Automated
Web Service Composition Methods', First
International Workshop on Semantic Web Services
and Web Process Composition (SWSWPC 2004), Jul.
2004, San Diego, CA, USA, pp. 43-54.
Kuter, U., et al., 2007, 'Conditionalization: Adapting
forward-chaining planners to partially observable
environments', Workshop on Planning and Execution
for Real-World Systems -- Principles and Practices for
Planning in Execution, Sept. 2007, Rhode Island,
U.S.A.
Lirong, Q, Zhongzhi, S & Fen, L, 2006, 'Context
Optimization of AI planning for Services
Composition', IEEE International Conference on e-
Business Engineering (ICEBE 2006), Oct. 2006, IEEE
Computer Society, Shanghai, China.
Marconi, A., et al., 'Enabling Adaptation of Pervasive
Flows: Built-in Contextual Adaptation.', 7th
International Joint Conference on Service Oriented
Computing (ICSOC- 2009), Nov. 2009, Stockholm,
Sweden
Merrick, I. & Wood, A 2000, 'Coordination with scopes',
ACM Symposium on Applied Computing-Volume 1,
Mar. 2000, Como, Italy.
Nau, D., Ghallab, M. & Traverso, P. 2004, Automated
Planning: Theory & Practice, Morgan Kaufmann
Publishers Inc., San Francisco, C. A, USA.
Nixon, L., et al., 2008, 'Tuplespace-based computing for
the semantic web: A survey of the state-of-the-art',
The Knowledge Engineering Review (2008),
Cambridge University Press, vol 23, no. 2, pp. 181--
212, .
Pistore, M., et al. 2004, 'Planning and Monitoring Web
Service Composition', Artificial Intelligence:
Methodology, Systems, and Applications, 11th
International Conference (AIMSA 2004), Sept. 2004,
Varna, Bulgaria.
Pistore, M, et al.2005, 'Automated Synthesis of Composite
BPEL4WS Web Services', IEEE International
Conference on Web Services (ICWS05), IEEE
Computer Society, Jul. 2005, Orlando, Florida, USA.
Rossi, D., Cabri, G. & Denti, E 2001, 'Tuple-based
technologies for coordination', Coordination of
Internet agents: models, technologies, and
applications, pp. 83-109, Springer-Verlag, London,
U.K.
Tolksdorf, R, Liebsch, F. & Nguyen, L., 2004,
'XMLSpaces.NET: An Extensible Tuplespace as XML
Middleware', 2nd International Workshop on.NET
Technologies,.NET Technologies’2004, May-Jun.
2004, Plzen, Czech Republic.
Tolksdorf, R., et al. 2005, 'Towards a tuplespace-based
middleware for the SemanticWeb', IEEE/WIC/ACM
International Conference on Web Intelligence
(WI2005), Sept. 2005, Compiegne University of
Technology, France.
Tsalgatidou, A., et al. 2008, 'Unified Discovery and
Composition of Heterogeneous Services: The
SODIUM Approach', in, At your service: An overview
of results of projects in the field of service engineering
of the IST programme, MIT Press Series on
Information Systems.
Vukovic, M. & Robinson, P. 2004, 'Adaptive, planning-
based, Web service composition for context
awareness', International Conference on Pervasive
Computing (Pervasive 2004), Apr. 2004, Vienna,
Austria.
Waldo, J. & Team, J. T. 2000, The Jini TM Specification,
2nd Edition, Addison-Wesley Professional.
Zeng, L., Lei, H & Chandramouli, B 2005, 'Semantic
Tuplespace', 3rd Internation Conference on Service
Orienteed Computing, (ICSOC-2005), Dec. 2005,
Amsterdam, Netherlands.
AN APPROACH TO DATA-DRIVEN ADAPTABLE SERVICE PROCESSES
145