SWAPT

Semantic Workﬂow Architecture for Petroleum Techniques

Nabil Belaid

1,2

, Yamine Ait-Ameur

and Jean-François Rainaud

Laboratory of Applied Computer Science (LISI), National Engineering School for Mechanics and Aerotechnics (ENSMA)

and University of Poitiers

1, avenue Clément Ader, BP 40109, 86961, Futuroscope, France

Technology, Informatics and Applied Mathematics Direction (DTIMA), Petroleum French Institute (IFP)

1 & 4, avenue de Bois-Préau, 92852 Rueil-Malmaison, France

Keywords:

Semantic workﬂows and web services, Annotations, Ontology-based modeling, Petroleum engineering.

Abstract:

In the petroleum industry, many engineering studies are conducted to evaluate the potential of the geological

structures to be exploited as hydrocarbon reservoirs. They are realized following series of complex workﬂows

composed by activities realized by geologists. Nowadays these workﬂows are build mainly according to the

experience gained by experts along their previous realizations. It is not possible to share this experience

between geologists by reusing and composing activities if a minimum of semantics is not applied to describe

them and the workﬂows that use them. The focus of our work is to evaluate the beneﬁt of using semantics

to make the geologist daily work easier. In this article, we ﬁrst explain how we can operate today without

semantics. Then, we enrich such complex workﬂows and the data they manipulate with semantic annotations

through ontology-based characterizations (Geological Data and Activities Ontologies). As future work, we

plan to use these annotations for a full architecture that would assist geologists in building their workﬂows.

1 INTRODUCTION

In the petroleum industry, several engineering studies

are conducted in order to deal with reservoir model-

ing. Indeed, seismic, geological and structural models

are set up. The geological modeling task is composed

of a series of multiple and complex processes or ac-

tivities. Workﬂows model these processes and their

composition. The orchestration is the action that sup-

ports the execution of these workﬂows.

Several work focus on automating or semi-

automating particular geological modeling activities.

However, the geologists knowhow and the com-

pounding activities are transfered into workﬂows

without any methodological rule. Our work focuses

on automating the activities orchestrations in the var-

ious processes of the geological modeling. The origi-

nality of our work lies on the semantics it brings in

different ways. This work suggests to use ontolo-

gies to characterize both activities, their composition

and the data they manipulate. Such a characterization

will help geologists in building their workﬂowsby ab-

stracting implementation details and focusing on their

semantics.

The remainder of this article is structured as fol-

lows. In section 2, we show the classical workﬂows

approach and its limitations and describe related work

aiming at bringing semantics to workﬂows. In section

3, we propose a semantic workﬂow approach. As a

case study, we present, in section 4 one of the work-

ﬂows of the petroleum ﬁeld and in section 5 the ge-

ological data it manipulates. The workﬂow activities

implementation is detailed in section 6. We ﬁnally

conclude and give some outlooks for future work.

2 FROM CLASSICAL TO

SEMANTIC WORKFLOWS

In classical workﬂow systems, activities manipulate

data model instances. No software vendor has imple-

mented such systems for geologists due to the fact that

depending on the objectives they establish, the geolo-

gists have to adapt their workﬂows.

101

Belaid N., Ait-Ameur Y. and Rainaud J.

SWAPT - Semantic Workﬂow Architecture for Petroleum Techniques.

DOI: 10.5220/0001840101010104

In Proceedings of the Fifth International Conference on Web Information Systems and Technologies (WEBIST 2009), page

ISBN: 978-989-8111-81-4

Because the activities in geology are strongly

linked to the technical descriptions of speciﬁc propri-

etary workﬂows, it would be difﬁcult to combine to-

gether atomic activities supplied by different vendors

in order to create new workﬂows even if we could iso-

late the atomic activities. To address this problem, we

suggest to enrich the geological data and the geologi-

cal activities with semantic annotations.

Our ﬁnal objective is to propose an architecture

that would enable geologists to perform tasks with a

minimum of geological knowledge instead of a tech-

nical computing knowledge.

Bringing semantics to workﬂows and Web ser-

vices is in the middle of the two hottest research ar-

eas associated with the Web: on the one hand, the

dynamic part through Web services and on the other

hand, the static part aiming at providing semantics to

knowledge and data.

The ﬁnal goal of Web services is to equip the Web

with distributed programs able to interact, to be re-

trieved automatically and to be composed into more

complex services. However, the interactions consen-

sus is not enough to allow unambiguous interaction

without Explicit semantics.

Ontologies have played their role in associating a

formal semantics to the Web service description. The

goal of OWL-S (Martin et al., 2004), is to enable

users and software agents to discover, invoke, com-

pose, and monitor Web resources. Another example

is SAWSDL. It is an upper layer added to WSDL and

recommended by the W3C (Farrell and Lausen, 2007)

that allows to deﬁne semantic terms used in WSDL by

referencing RDF-based ontologies. Many work focus

on the automatic discovery of Web services. For ex-

ample, (Bernstein and Klein, 2002) uses ontologies of

processes to describe the services behavior and deﬁne

a Process Query Language. Finally, many other work

focus on the automatic composition of Web services.

For example, (Hendler et al., 2003) exploit annotation

tools for services and scheduling in order to be able to

compose services to create predeﬁned functionalities.

3 SEMANTIC WORKFLOW

ARCHITECTURE (SWA)

The main idea behind our SWA is to try to bring se-

mantics to every part that composes workﬂows. Cur-

rent work are concerned with interpreting the data by

annotating them with ontological concepts (see Sec-

tion 5).The ﬁrst part of our proposal for a SWA uses

this enriched data as the exchanged elements of the

workﬂows supported by the architecture. The second

part consists in providing a semantic characterization

to the workﬂows themselves.

Figure 1 shows the layers of our SWA and the an-

notations that link them.

Figure 1: Semantic Workﬂow Architecture.

• Data Models (DM)

The DM instances are the basic units manipulated

in our architecture. They can be basic types like

integers or strings, complex types or even ﬁles.

They are instances of DM concepts which repre-

sent types or formats. Classical workﬂows manip-

ulate instances of DM.

• Data Ontology (DO)

The DO instances are instances of DO concepts.

They can interpret DM instances through annota-

tions (semantic relations) which link pairs of in-

stances (an instance of DM is semantically anno-

tated by an instance of DO).

• Activities Models (AM)

Currently, activities use as input and output DM

instances. Different activities manipulate DM In-

stances which are in accordance with different and

speciﬁc formats that are interpreted as being the

same DO instance. Moreover, the same format

may have different meanings according to which

activity manipulates it. Different DO instances

can interpret different DM instances which are de-

ﬁned in a common format.

• Semantic Workﬂows (AO)

AO represents semantic Web services and each

Web service implements a speciﬁc geological

AM. The semantic services annotate one or more

AM. One or more implementations of Web ser-

vices may correspond to a semantic Web service.

Likewise, to a semantic workﬂow can correspond

one or more implementations of workﬂows and

Web services. AO can be semantic Web services

(atomic activities) or semantic workﬂows (com-

posite activities).

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

102

4 CASE STUDY: GEOLOGICAL

MODELING WORKFLOWS

The geological modeling with the view to securely

store CO2 can be seen as a high level workﬂow.

This workﬂowaims at transforming a "3D-image" ob-

tained from a seismic exploration to a ﬂow simulation

forecast. This enables to predict how any ﬂuid, and in

our case CO2, would propagate in a studied prospect.

The seismic interpretation is the up-stream of the

geological modeling. Its objective is to interpret a

3D-image represented by a seismic cube to create a

structural model and recognize the geological struc-

ture elements like horizons, faults or channels.

The seismic interpretation is a complex process

and can be modeled as a workﬂow. It manipulates

data, such as 3D-images (SEG-Y Files) represent-

ing seismic cubes or ﬁles of coordinates points (XYZ

Files) representing horizons, reﬂectors etc. After a se-

quence of interpretation activities, a structural model

is obtained. Figure 2 zooms on a candidate seismic

interpretation workﬂow.

Figure 2: Up-stream fragment of a seismic interpretation

workﬂow.

5 GEOLOGICAL DATA (GD)

The geological modeling workﬂows manipulate geo-

logical data which are instances of DM concepts. One

of the objectives of our work is to make it possible for

the activities to manipulate DO instances.

In (Mastella et al., 2008), the idea is to create an

ontology for each geological ﬁeld; the seismic ontol-

ogy for instance. Then, an annotation model able to

link the data to the ontology instances is set up. Fig-

ure 3 shows the relation between the geological data

models (A), the geological data ontology (B) and the

annotations model (C).

• Geological data models (GDM)

In the geology ﬁeld, the geologist manipulates

GDM such as SEG-Y Files that represent seismic

cubes (see Figure 3.A).

Figure 3: Fragment of the geological data ontology, the data

models and the annotations model.

• Geological Data Ontology (GDO)

In the context of the project e-Wok Hub

that

deals with the storage of CO2 by ontologies-based

modeling

, experts have deﬁned a consensual and

shared geological ontology (see Figure 3.B).

• Geological Data Annotation Model (GDA)

An annotation can be for example Documen-

tAnnotation when the annotation refers to GDM

which are ﬁles (see Figure 3.C).

When the geologist creates an annotation, it refer-

ences an instance of an OntologicalConcept and

an instance of a DataModelClass. As an example,

Figure 4 shows an instance of a GDM instance of

XYZ File (whose name is "reﬂect3D_0047.xyz")

that is annotated by a GDO instance of the con-

cept Reﬂector (whose URI is "r1").

Figure 4: Example of a data annotated by an ontological

instance.

6 SEMANTIC WORKFLOWS:

IMPLEMENTATIONS

We have created a model and an ontology for the ac-

tivities. We then created an annotation model to link

the model and the ontology instances. Figure 5 shows

the relation between the geological activites models

(A), the geological activites ontology (B) and the an-

notations model (C).

A semantic activity ontology is the ﬁrst part to-

ward an architecture that proposes workﬂows and as-

sists geologists in their geological modeling tasks.

http://www-sop.inria.fr/edelweiss/projects/ewok/

ANR project involving the following partners:

LISI/ENSMA, IFP, BRGM, INRIA, EADS, Paris Mines

School and the CRITT Informatics.

SWAPT - Semantic Workflow Architecture for Petroleum Techniques

103

Figure 5: Fragment of the geological activities model, the

geological activities ontology and the annotation model.

• Geological Activities Models (GAM)

In geology modeling workﬂows, the activities are

designed independently one from the others. As

a consequence, a set of heterogeneous activities

and models, that manupulate instances of GDM,

are created (see Figure 5.A).

• Geological Activities Ontology (GAO)

As a second step, we have then created an on-

tology of semantic activities (Web services) that

would enable a semantic search over the Web ser-

vices (see Figure 5.B). Indeed, when many Web

services implement the same activities, a unique

GAO concept corresponds to the given action. It

is possible then to retrieve one or more (in the

case of multiple Web services) WSDL descrip-

tions and/or one or more workﬂows (e.g. BPEL).

• Geological Activities Annotation (GAA)

GAM instances are annotated by GAO instances

in the same way GDM instances are annotated by

GDO instances.

For example, the semantic activity FaultsDe-

tectAct annotates both FaultsDetectAct Wf1 and

FaultsDetectAct Wf2 (see Figure 6 for the anno-

tation example).

Figure 6: Example of an activity annotation.

7 CONCLUSIONS

We have described our proposal for an approach that

intends to assist geologists in building their workﬂows

by adding semantic annotations to the activities and

to the data they manipulate through ontologies char-

acterization.

The work presented in this article was the ﬁrst part

toward a full architecture supporting semantic geolog-

ical workﬂows. In future work, we turn to the persis-

tence of atomic and composite activities executions.

Recently, several systems were proposed to store

in the same database the data and the ontologies de-

scribing them: ontology-based databases (OBDBs).

OntoDB is one of them (Dehainsala et al., 2007). One

of the advantages is the possiblity of querying the

databases at the ontology level (Jean et al., 2006).

Thus, we plan to store GDO, GDM and the anno-

tation instances in the same OntoDB. We intend to es-

tablish a meta-model of activities and record all GAO,

GAM and annotations instances in the same previous

OntoDB which will enable complex semantic queries.

REFERENCES

Bernstein, A. and Klein, M. (2002). Discovering services:

Towards high precision service retrieval. In CaiSE

workshop on Web Services, e-Business, and the Se-

mantic Web: Foundations, Models, Architecture, En-

gineering and Applications. Toronto, Canada.

Dehainsala, H., Pierra, G., and Bellatreche, L. (2007).

Ontodb : An ontology-based database for data in-

tensive applications. 12th Int. Conf. on Database

Systems for Advanced Applications (DASFAA’07),

LNCS. Springer, p. 497-508.

Farrell, J. and Lausen, H. (2007). Semantic annotations for

wsdl and xml schema. Technical report, W3C Recom-

mendation.

Hendler, J., Nau, D., Parsia, B., Sirin, E., and Wu, D.

(2003). Automating daml-s web services composition

using shop2. ISWC’03.

Jean, S., Pierra, G., and Ait-Ameur, Y.(2006). Querying on-

tology based database using ontoql (an ontology query

language). Ontologies, Databases, and Applications

of Semantics (ODBASE’06), p. 704-721.

Martin, D., Burstein, M., Hobbs, J., Lassila, O., McDer-

mott, D., McIlraith, S., Narayanan, S., Paolucci, M.,

Parsia, B., Payne, T., Sirin, E., Srinivasan, N., and

Sycara, K. (2004). Owl-s: Semantic markup for web

services. Technical Report.

Mastella, L. S., Ait-Ameur, Y., Perrin, M., and Rainaud,

J.-F. (2008). Ontology-based model annotation of

heterogeneous geological representations. in 4th Int.

Conf. on Web Information Systems and Technologies

(WEBIST). Madeira, Portugal.

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

104