SW-ONTOLOGY

A Proposal for Semantic Modeling of a Scientiﬁc Workﬂow Management System

Wander Gaspar, Laryssa Silva, Regina Braga and Fernanda Campos

Computational Modeling Master Program, Federal University of Juiz de Fora, Juiz de Fora, Brazil

Keywords:

Ontology, e-Science, Scientiﬁc workﬂow, Semantic.

Abstract:

The execution of scientiﬁc experiments based on computer simulations constitutes an important contribution to

scientiﬁc community. In this sense, the implementation of a scientiﬁc workﬂow can be automated by Scientiﬁc

Workﬂow Management Systems, which goal is to provide the orchestration of all processes involved. It aims to

capture the semantic related to the implementation of scientiﬁc workﬂows using ontologies that could capture

the knowledge involved in these processes. Speciﬁcally, we present a prototype of an ontology based on

a design pattern called Model View for the representation of knowledge in scientiﬁc workﬂow management

systems.

1 INTRODUCTION

We can consider that scientiﬁc research is based on

three pillars: theory, experimentation and computa-

tional resources. The use of these resources helps the

sharing of data, tools and services, and allows the sys-

tematic reuse of experiments. Considering this sce-

nario, researches need an infrastructure that allows the

design, reuse, annotation, validation, sharing and doc-

umentation of the work done by scientists (Barga and

Digiampietri, 2008).

The implementation of experiments based on

computer simulations constitutes an important con-

tribution to the scientiﬁc community. In most cases,

current practice is to implement a set of software and

scripts. This procedure has proved insufﬁcient to ad-

equately handle the inherent complexity of the prob-

lems with which scientists have come across. In this

context, it was deﬁned the e-Science term, the sci-

ence that has been largely supported by simulation

and computational infrastructure, based on techniques

like scientiﬁc workﬂows and web services.

In e-Science, a major goal is the creation and use

of processes that simulates experiments, analyzes data

and discovers knowledge, using a wide range of com-

puting resources (Wroe et al., 2007). Technologies

such as ontologies and semantic web services can be

used as the basis for the composition of an infrastruc-

ture to support e-Science (Silva et al., 2009). This pa-

per considers this scenario, presenting the use of on-

tologies and the composition of semantic web servi-

ces in different subdomains in the context of e-

Science. Speciﬁcally, we present the SW-Ontology,

an ontology for the knowledge representation in Sci-

entiﬁc Workﬂow Management Systems (SWfMS).

The article is organized as follows. Section 2

presents the background for the research, includ-

ing SWfMS and the use of ontologies for knowl-

edge representation. Section 3 describes some related

works. Section 4 presents the SW-Ontology, a se-

mantic model based on ontologies for scientiﬁc work-

ﬂows. Section 5 details the use of SW-Ontology for

the composition of scientiﬁc workﬂows, and ﬁnally,

Section 6 presents ﬁnal considerations and suggests

some future researches.

2 CONCEPTS AND RELATED

WORKS

In a historical perspective, a scientiﬁc experiment is

one of the tools used by researchers to support the

formulation of new theories. In this context, a sci-

entiﬁc workﬂow represents the orchestration of pro-

cesses that handle data in order to build a simulation.

2.1 Scientiﬁc Workﬂow Management

Systems

A workﬂow could be deﬁned as a description of a re-

producible process consisting of a set of interrelated

115

Gaspar W., Silva L., Braga R. and Campos F. (2010).

SW-ONTOLOGY - A Proposal for Semantic Modeling of a Scientiﬁc Workﬂow Management System.

In Proceedings of the 12th International Conference on Enterprise Information Systems - Databases and Information Systems Integration, pages

115-120

DOI: 10.5220/0002865601150120

 SciTePress

tasks (Menager and Lacroix, 2006). The execution of

a scientiﬁc workﬂow can be automated using com-

putational tools called Scientiﬁc Workﬂow Manage-

ment Systems, whose goal is to orchestrate the design,

management and implementation of scientiﬁc exper-

iments. The present study intend to capture the se-

mantics involved on the scientiﬁc workﬂows orches-

tration in SWfMS, considering the design process and

the implementation of workﬂows, using ontologies to

capture the knowledge involved in these processes.

2.2 Ontologies

The word ontology comes from the Greek ontos (be)

+ logos (word). In Philosophy, it is the science of

what is, of the types and structures of objects, proper-

ties, events, processes and relations in every domain.

In this context, the purpose of an ontology is to pro-

vide categorization systems to organize the reality.

Considering Semantic Web, the deﬁnition more often

cited in the literature is that an ontology is a formal

and explicit speciﬁcation of a shared conceptualiza-

tion (Gruber, 1993).

From the 1990s, several languages were proposed

for the representation of ontologies. At the same time,

the rapid expansion of the Internet led to the emer-

gence of lightweight markup languages to support and

at the same time explore the World Wide Web charac-

teristics.

In this context, the Word Wide Web Consor-

tium (W3C) launched and formally recommended as

a standard the Web Ontology language (OWL), de-

signed to meet the requirements of the Semantic Web.

Prot

e, an editor of ontologies and knowledge bases

supports OWL (Horridge, 2009). In addition, a wide

range of inferences and OWL validation machines are

available, such as Pellet (Sirin et al., 2007) e FaCT++

(Tsarkov and Horrocks, 2006), and semantic Web

frameworks supporting OWL such as Jena (Hewlett-

Packard, 2009).

2.3 Related Work

There are several works related to semantic represen-

tation in scientiﬁc workﬂows. The main contribution

of our work is the use of SW-Ontology in shaping the

composition of scientiﬁc workﬂows in the context of

e-Science, helping the scientist in modeling the more

suitable scientiﬁc workﬂow for the experiment to be

executed.

The OWL ontology myGrid (Wolstencroft et al.,

2007) was modeled for discovering and composition

of web services in Bioinformatics domain using the

Taverna SWfMS (Oinn et al., 2004) with semantic an-

notations, where you can use inference to ﬁnd com-

mon ancestors to the activities of workﬂows. The

myGrid ontology models knowledge into scientiﬁc

workﬂows based on super-classes algorithm, date,

metadata, task, data resource, ﬁle formats and ser-

vice. One of the drawbacks of this approach is that

it does not present a clear separation between classes

related to modeling and visualization for the domain

addressed. SW-Ontology seeks to extend the domain

of scientiﬁc workﬂows, including a clear separation

between view and modeling classes, with the possi-

bility of expanding the scope of ontology for several

similar software systems.

In (Fox et al., 2009) is presented a semantic data

framework that models an OWL-DL ontology for the

representation of knowledge in the sub-domain of

Physics related to the Sun and the Earth, describing

concepts, relations and attributes of physical mag-

nitudes. The ontology is divided into main classes

Instrument, Observatory, Operating Mode, Parame-

ter, Coordinate and Data Archive. In (Oliveira et al.,

2009) is presented an ontology for the semantic mod-

eling of scientiﬁc workﬂows related to oil exploration

in deep waters. The ontology was used to deﬁne

some semantic concepts in order to provide support

for workﬂow composition. A case study is discussed

and, according to the authors, the results reinforce the

beneﬁts of semantic support during the manual chain-

ing of processes and subworkﬂows. In both works,

the emphasis is on classes related to models. None of

them presents classes related to the implementation of

a scientiﬁc workﬂow.

2.4 Scope of the Work

SW-Ontology aims to describe the knowledge rep-

resented in scientiﬁc workﬂows, emphasizing such

modeling in the context of SWfMS Vistrails. In ad-

dition, SW-Ontology tries to incorporate to Vistrails

semantic modeling facilities such as resources for

queries and analysis of data provenance.

Vistrails represents a scientiﬁc workﬂow as an

acyclic graph. This SWfMS gives great emphasis on

data and process provenance and allows comparisons

of results to generate complex views. The choice

for this SWfMS was based on our group interest in

data provenance and Vistrails has an interesting data

provenance mechanism. Besides, Vistrails is related

with other developing works at the Research Center

on Software Quality (NPQS) of Federal University of

Juiz de Fora (UFJF) that uses this environment for sci-

entiﬁc workﬂows design and implementation. Cur-

rently, the NPQS focuses on the provision of an in-

frastructure for e-Science, named ASOW-Science

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

116

(Matos et al., 2009) that includes technologies like on-

tologies, components and agents, with applications in

Bioinformatics, Agriculture and Education.

The SW-Ontology was developed using OWL-

DL, which ensures high capacity of expressiveness

and the inference computability in a ﬁnite time. The

adoption of OWL-DL is reinforced by the current sta-

tus of the language that is recommended by W3C

Consortium as part of a set of technologies for the

development of Semantic Web. The prototype im-

plementation was done in the Prot

e, an environ-

ment for creating and editing ontologies and knowl-

edge bases (Horridge, 2009).

3 SEMANTIC MODELING OF A

SWfMS

For SW-Ontology development, we chose to use

a design pattern named Model-View (MV), which

is a derivation of the design pattern Model-View-

Controller (MVC). The purpose of the MVC pattern

arose from the increase growing of software devel-

opment complexity. In this context, it is essential

to separate the data (model), the layout (view) and

the control (controller). Thus, the implementation of

the MVC design pattern allows that changes in layout

does not affect the data and them the data can be reor-

ganized without signiﬁcant changes in layout. As part

of this work, the MV design pattern is used in order to

separate the knowledge on the semantics of scientiﬁc

workﬂows design and implementation (model) from

the human-computer interaction performed by using

the SWfMS graphical user interface (view).

The use of MV is also important on the categoriza-

tion of classes, because it clearly deﬁnes the scope of

the model. It is still possible to glimpse the deﬁni-

tion of different ontology view for a single ontology

model. Speciﬁcally in the context of this work, it is

possible to build view classes for the modeling do-

main from other SWfMS such as Taverna (Oinn et al.,

2004) and Kepler (Lud

ascher et al., 2006), to name a

few.

In SW-Ontology, the subclasses model has been

divided in two hierarchies, i.e., as shown in Fig-

ure 1, the Model DomainEntity hierarchy for classes

of model and the ViewEntity for classes of view.

Figure 1: SW-Ontology main classes.

The ModelDomainEntity hierarchy is shown in

Figure 2. In Vistrails, a workﬂow can be deﬁned as

a set of interconnected modules (class Module). The

links between modules are made from input and out-

put ports (class Port) and the actions taken by a mod-

ule are processed by methods (class Method) which

may receive parameters (class Parameter) for its im-

plementation. The various modules provided by Vis-

trails or by third parties are categorized into packages

(class Package). Vistrails also allows access to Web

services (class WebService) in the workﬂows compo-

sition. Finally, considering that this is an environment

for collaborative scientiﬁc exploration, it allows users

to store not only stand-alone ﬁles but also data from

relational databases (class Workﬂow StorageMode).

Figure 2: Model class hierarchy from ModelDomainEntity.

The view class ViewEntity describes the knowl-

edge of Vistrails users interaction environment (Fig-

ure 3). BuilderViewEntity represents the set of graph-

ical interfaces used for composition, execution and

query within scientiﬁc work-ﬂows. SpreadsheetView

models the knowledge related to results visualization

and exploitation.

Figure 3: View class hierarchy from ViewEntity.

The current version of SW-Ontology contains 68

classes, 38 object properties (properties that indicate

a relationship between two classes) and 15 data prop-

erties (properties that indicate a relationship between

instances of classes and literals expressed in RDF or

XML Schema data types). All classes have annota-

tions comment type, whose goal is to provide a de-

scription of the knowledge modeled.

Figure 4 shows the representation of the model

SW-ONTOLOGY - A Proposal for Semantic Modeling of a Scientific Workflow Management System

117

class Workﬂow on Prot

e, deﬁned as a set of inter-

connected modules in order to model a workﬂow. One

can observe that the adoption of the MV design pat-

tern allows that we can explicitly present the super

class relationship between Model DomainEntity and

Workﬂow class.

Figure 4: Modeling the model class Workﬂow on Prot

Figure 5 shows the view class QueryInterface,

which models the knowledge related to the deﬁnition

of queries such as query-by-example in a predeﬁned

workﬂow that can locate and display graphically sub-

workﬂows or modules that meet the query performed.

Figure 5: Modeling the model class QueryInterface.

As an example of some restriction modeled in

SW-Ontology, we can see that the restriction has-

Workﬂow some Workﬂow relates an individual from

model class Workﬂow to the view class QueryInter-

face according to the property has-Workﬂow. This re-

striction may be interpreted as: it is necessary the ex-

istence of a previously constructed workﬂow in order

to run a query in the graphical interface. Thus, the

use of restrictions conﬁgures itself in a mechanism

capable of providing the connection between model

and view classes.

The Pellet (Sirin et al., 2007) was used for the

class hierarchy inference and SW-Ontology consis-

tency check. The use of an inference engine al-

lows you to extract new knowledge from the ontol-

ogy model built. Figure 6 illustrates results presented

for UserDeﬁnedPackage, a ModelDomainEntity sub-

class, after SW-Ontology classiﬁcation by Pellet. You

can verify that the reasoner has derived the restric-

tions HasModule some ModulesPanel and boolean

isEnabled exactly 1 from the class hierarchy built by

the classiﬁer and their descriptions.

Figure 6: Modeling the model class UserDeﬁnedPackage.

4 COMPOSITION OF

SCIENTIFIC WORKFLOWS:

USE OF SW-ONTOLOGY

ASOW-Science is a framework based on semantic

web services to compose workﬂows in a scenario of

e-Science. Can be understood as the speciﬁcation and

development of an infrastructure whose purpose is to

provide computational support for researchers who

want to share experiments and results in a given ap-

plication domain (Silva et al., 2009). Speciﬁcally,

ASOW-Science manages the storage of ontologies

and semantic web services in distributed repositories,

and provide resources to the scientist to perform se-

mantic queries to the database. Furthermore, it is able

to make an automatic analysis of the services discov-

ered, in order to obtain possible compositions that can

be used to design workﬂows in a SWfMS.

Figure 7 represents the ASOW-Science layers.

The framework has two components: a client compo-

nent that invokes the service, and a middleware com-

ponent. The middleware consists of four layers:

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

118

• the Backend Layer contains the database for stor-

age and query ontologies of the domain and se-

mantic annotations of Web services;

• the Semantic Layer is intended to manage the pro-

cesses of storage and query ontologies and seman-

tic annotations of services;

• the Search Layer performs the semantic search

and discovery of services according to the scien-

tist speciﬁcations. The information provided by

scientist and obtained by inferences is used to per-

form semantic search in the repositories to ﬁnd

web services semantically compatible with each

task in the workﬂow. To ﬁnd services, semantic

descriptions of services available in repositories

are analyzed and compared with the semantic data

related to each task;

• The Application Layer is responsible for model-

ing candidate compositions of scientiﬁc workﬂow

using the services discovered by the search layer.

Figure 7: The proposed framework architecture.

The integration of SW-Ontology with ASOW-

Science aims to provide scientists with a tool that fa-

cilitates the orchestration of a computer simulation in

the context of e-Science. After selecting the ontology

in the framework, the researcher can view the classes

and restrictions available, along with their semantic

descriptions. The ultimate goal of ASOW-Science is

to provide the scientists a search engine for semantic

Web services that exist in the framework repositories

and capable of performing the tasks selected from the

ontology.

To test ideas, we built a prototype that executes

a scientiﬁc workﬂow related to cell models speciﬁed

in CellML language (Matos et al., 2010). This work-

ﬂow, related to the ﬁeld of cardiac electrophysiology,

and developed at the UFJF Laboratory of Compu-

tational Physiology and High-Performance Comput-

ing, has several variations in terms of tools to add to

it. Thus, the proposed framework could be used to

semantically discover Web services which are more

suitable for the workﬂow implementation.

In this context, SW-Ontology and CELO (the do-

main ontology) are used together by the ASOW-

Science framework to provide the relationships

among the types of components selected by the re-

searcher and build a scientiﬁc workﬂow capable of

meeting the requirements.

In Figure 8 we have the execution schema of this

scientiﬁc workﬂow. Component 1 gets the system

date and current time, concatenate them in a sequence

of characters and create the ﬁles necessary to exe-

cute Component 2, using the sequence of characters

formed from the date of the system in their names.

Component 2 encapsulates a compiler that generates

an executable C code from a CellML model (Beard

et al., 2009), and executes the code to obtain output

data — the solution of an ordinary differential equa-

tion (ODE). Component 3 encapsulates a tool capa-

ble of generating a graph from a text ﬁle containing

ordered pairs. This component receives as input the

output ﬁle of Component 2 and creates a ﬁle to the

generated graph. Its output is the URL of the graph

ﬁle. Finally, Component 4 displays the graphical so-

lution of the ODE.

Figure 8: Cardiac electrophysiology workﬂow schema.

Considering the workﬂow shown in Figure 8,

classes of SW-Ontology as Workﬂow, Module and

ConcatenateStringModule, among others, represent

the related knowledge which constitutes the basis for

the composition of a scientiﬁc workﬂow within the

prototype.

Currently, there is a prototype of the proposed ar-

chitecture, developed as a Web service. This proto-

type allows that the scientist choose terms from SW-

Ontology and CELO ontology and links these terms

to tasks that must compose the workﬂow. Analyz-

ing the semantic annotation of Semantic Web services

that are stored in a repository, these terms could be

used to discover services that best ﬁt the model pro-

vided by the user.

SW-ONTOLOGY - A Proposal for Semantic Modeling of a Scientific Workflow Management System

119

5 FINAL CONSIDERATIONS

Researches in e-Science have gained increase rele-

vance. However, there are few studies in the ﬁeld

of Software Engineering focused on the topic and, in

particular, as in the use of ontologies in the e-Science

researches (Palazzi et al., 2009).

This paper proposes a model of semantic descrip-

tion based on OWL-DL ontologies to describe the

knowledge related to scientiﬁc workﬂow design and

implementation. Using the MV design pattern, an

ontology was built, separating the concepts related

to the semantic of scientiﬁc workﬂows orchestration

(model) and from the human-computer interaction

as of the computing environment graphical user in-

terface used (view). Currentlly, we are using SW-

Ontology in a huge project related to scientiﬁc work-

ﬂow speciﬁcation in the human diseases domain.

REFERENCES

Barga, R. S. and Digiampietri, L. A. (2008). Auto-

matic capture and efﬁcient storage of e-science exper-

iment provenance. Concurr. Comput. : Pract. Exper.,

20(5):419–429.

Beard, D. A., Britten, R., Cooling, M. T., Garny, A., Hal-

stead, M. D. B., Hunter, P. J., Lawson, J., Lloyd,

C. M., Marsh, J., Miller, A., Nickerson, D. P., Nielsen,

P. M. F., Nomura, T., Subramanium, S., Wimalaratne,

S. M., and Yu, T. (2009). Cellml metadata standards,

associated tools and repositories. Physical and Engi-

neering Sciences, 367(1895):1845–1867.

Fox, P., McGuinness, D. L., Cinquini, L., West, P., Gar-

cia, J., Benedict, J. L., and Middleton, D. (2009).

Ontology-supported scientiﬁc data frameworks: The

virtual solar-terrestrial observatory experience. Com-

put. Geosci., 35(4):724–738.

Gruber, T. R. (1993). A translation approach to portable on-

tology speciﬁcations. Knowl. Acquis., 5(2):199–220.

Hewlett-Packard (2009). Jena semantic web framework.

Horridge, M. (2009). Prot

e owl tutorial. Technical Report

v.1.2, The University of Manchester.

Lud

ascher, B., Altintas, I., Berkley, C., Higgins, D., Jaeger,

E., Jones, M., Lee, E. A., Tao, J., and Zhao, Y. (2006).

Scientiﬁc workﬂow management and the kepler sys-

tem. Concurr. Comput. : Pract. Exper., 18(10):1039–

1065.

Matos, E. E., Mendes, L. F., Campos, F., and Braga, R.

(2009). Asow-science: a service oriented framework

to support e-science applications. In IRI’09: Proceed-

ings of the 10th IEEE international conference on In-

formation Reuse & Integration, pages 53–56, Piscat-

away, NJ, USA. IEEE Press.

Matos, E. E. E., Campos, F., Braga, R., and Palazzi, D.

(2010). Celows: an ontology based framework for

the provision of semantic web services related to bi-

ological models. Journal of biomedical informatics,

43(1):125–136.

Menager, H. and Lacroix, Z. (2006). A workﬂow engine for

the execution of scientiﬁc protocols. In ICDEW ’06:

Proceedings of the 22nd International Conference on

Data Engineering Workshops, page 68, Washington,

DC, USA. IEEE Computer Society.

Oinn, T., Addis, M., Ferris, J., Marvin, D., Carver, T.,

Pocock, M. R., and Wipat, A. (2004). Taverna: A tool

for the composition and enactment of bioinformatics

workﬂows. Bioinformatics, 20:3045–3054.

Oliveira, D. d., Cunha, L., Tomaz, L., Pereira, V., and Mat-

toso, M. (2009). Using ontologies to support deep wa-

ter oil exploration scientiﬁc workﬂows. In SERVICES

’09: Proceedings of the 2009 Congress on Services -

I, pages 364–367, Washington, DC, USA. IEEE Com-

puter Society.

Palazzi, D., Silva, L., Mendes, L., Gaspar, W., Matos, E.,

Campos, F., and Braga, R. (2009). Using ontologies

in e-science projects (in portuguese). In II Seminar on

Ontology Research in Brazil.

Silva, L., Campos, F., and Braga, R. (2009). A framework

for semantic composition of scientiﬁc workﬂows. In

IADIS’09: Proceedings of the International Confer-

ence WWW/Internet.

Sirin, E., Parsia, B., Grau, B. C., Kalyanpur, A., and Katz,

Y. (2007). Pellet: A practical owl-dl reasoner. Web

Semantics, 5(2):51–53.

Tsarkov, D. and Horrocks, I. (2006). Fact++ description

logic reasoner: System description. In Proc. of the Int.

Joint Conf. on Automated Reasoning (IJCAR 2006),

volume 4130 of Lecture Notes in Artiﬁcial Intelli-

gence, pages 292–297. Springer.

Wolstencroft, K., Alper, P., Hull, D., Wroe, C., Lord, P. W.,

Stevens, R. D., and Goble, C. A. (2007). The my-

grid ontology: bioinformatics service discovery. Int.

J. Bioinformatics Res. Appl., 3(3):303–325.

Wroe, C., Goble, C., Goderis, A., Lord, P., Miles, S., Pa-

pay, J., Alper, P., and Moreau, L. (2007). Recycling

workﬂows and services through discovery and reuse:

Research articles. Concurr. Comput. : Pract. Exper.,

19(2):181–194.

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

120