TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED

FLEXIBLE COLLABORATIVE SYSTEMS

A Basic Framework Applied to Medical Research

Jonas Schulte, Thorsten Hampel

Heinz Nixdorf Institute, University of Paderborn, Paderborn, Germany

Konrad Stark, Johann Eder, Erich Schikuta

Department of Knowledge and Business Engineering, University of Vienna, Vienna, Austria

Keywords:

SOA, Framework, CSCW, Wasabi, Medical Research Environment.

Abstract:

Collaborative systems have to support speciﬁc functionalities in order to be useful for special ﬁelds of appli-

cation and to fulﬁl those requirements. In this paper we introduce the Wasabi framework for collaborative

systems, which is characterised by ﬂexibility and adaptability. The framework implements a service oriented

architecture and integrates different persistence layers. The requirement analysis for the Wasabi CSCW system

is presented in the context of a collaborative environment for medical research which has strict requirements

concerning data integrity. This paper shows the results of the requirement analysis and how these are imple-

mented in the Wasabi architecture.

1 INTRODUCTION

In arbitrary ﬁelds of application all types of coop-

eration follow a similar pattern: A group of indi-

viduals share data and knowledge within a certain

context, add supplementary data, and deduce new

knowledge by applying functions. Their results are

stored for future collaboration or for other interest

groups. Contemporary Computer Supported Coop-

erative Work (CSCW) systems either support only a

subset of functionalities useful for the users’ daily

work or have restricted ﬁelds of application. Hence,

most systems are inﬂexible and adapting a system for

a new scope of application is a challenging task. De-

veloping an infrastructure for CSCW systems char-

acterised by ﬂexibility and adaptability is desirable,

since each scope of application has speciﬁc require-

ments resulting from their ways and methods of col-

laboration. Speciﬁc requirements identiﬁed for a ﬁeld

of application can be deﬁned for functionalities of the

CSCW system which support users in terms of exe-

cuting their daily workﬂows. Furthermore the CSCW

system should be able to handle arbitrary data types

and data repositories. In particular, handling various

persistence layers enables to bind remote repositories

of different types or with different characteristics (se-

curity, repository management, etc.). The Wasabi in-

frastructure presented in this paper aims to build a

next generation framework for CSCW systems by fo-

cussing on ﬂexible service integration – in this con-

text of a Service Oriented Architecture (SOA) – and

object oriented data integration (integration of persis-

tency layers).

The beneﬁts for such a ﬂexible infrastructure can

be obviously seen in connection with a collaborative

system for medical research which is used in this pa-

per as an example ﬁeld of application for our frame-

work. The collaborative environment for medical re-

search is needed in the context of the biobank ini-

tiative GATiB (Genome Austria Tissue Bank) which

is part of the Austrian Genome Program (GEN-AU,

2007). GATiB aims at the establishment of a tissue

bank (BIOBANK, 2007) which builds on a collection

of diseased and corresponding normal tissues repre-

senting all diseases at the natural frequency of oc-

currence from a non-selected central European pop-

ulation of more than 700.000 patients. Major em-

phasis is placed on annotation of archival tissue with

comprehensive clinical data including follow-up data

(medication, therapy, resection). A more detailed de-

scription of the biobank initiative is given in (Asslaber

et al., 2007).

In the context of collaborative medical research

various conceptual aspects have to be considered:

Which individuals are involved in collaboration activ-

232

Schulte J., Hampel T., Stark K., Eder J. and Schikuta E. (2008).

TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED FLEXIBLE COLLABORATIVE SYSTEMS - A Basic Framework Applied to Medical

Research.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - DISI, pages 232-239

DOI: 10.5220/0001696302320239

 SciTePress

ities, which sources are accessed for which purpose,

and what are the prevailing results? Collaboration in

the medical research ﬁeld is characterised by a high

complexity and high variation of collaborative situa-

tions. Data is distributed over several institutes and

underlays various restrictions in it’s accessibility for

different persons (roles). Access to patient data is for

example highly limited to the context or role of ap-

plication (e.g. patient context data or disease context

data). Data is collected, restructured, analysed, and

shared with other persons in different settings.

Even if collaboration activities in arbitrary ﬁelds

of application are very similar concerning the user’s

activities, the medical research scenario has very strict

requirements because of the sensitive data. (Stark

et al., 2008) presents several scenarios and a practi-

cal example for the usage of the cooperative system in

GATiB. For other communities/applications less strict

differentiations exist. However, in its characteristic as

a collaborative knowledge management system, our

Wasabi-CSCW infrastructure applied to GATiB can

be easily used as an e-learning platform or for other

ﬁelds. Finally the ﬂexible conﬁguration of access

rights and the architecture’s ability to deﬁne contex-

tual views on the presented data allows to use it as a

powerful collaborative learning platform.

The paper is structured as follows. To describe

the background of the requirements for a CSCW sys-

tem, which is usable for medical research, Section 2

presents example collaboration scenarios and sum-

marises the main requirements for such a system.

An overview of the Wasabi-CSCW system applied to

GATiB is described in Section 3. Details according

to the fulﬁlment of the requirements are described in

Section 4. Our paper ends with an overview of related

work in Section 5, followed by some conclusions.

2 EXAMPLE USAGE AND

REQUIREMENTS

Collaboration in the GATiB project focuses on both

the medical research as well as routine activities of

medical scientists and supporting staff. Data locally

distributed over institutes and research groups is ac-

cessed in manifold ways and for various purposes.

Further it is collected, restructured, augmented, anal-

ysed, and shared within a certain group of interested

researchers. In the following an example scenario is

given.

Due to a cooperation between a hospital and a

pharmaceutical company a group of suitable human-

tissue donors is to be identiﬁed to support a medi-

cation discovery study. Therefore, pathological di-

agnoses, survival data, and tissue images of patients

that have signed a formal consent are required. Af-

ter searching and structuring information, conﬁden-

tial patient data has to be protected since an external

organisation is involved in the study. Hence, identi-

fying attributes (name, day of birth) are eliminated

and quasi-identifying attributes are k-anonymised

(P. Samarati, 1998; Stark et al., 2006). Life style data

is included by ﬁlling out questionnaires. Further, a tis-

sue microarray of the relevant cases is made in order

to test candidate tumour markers. Further, the results

should be made available to other research groups.

2.0.1 Requirements

The scenario demonstrates the diversity of collabora-

tion types that may occur. A cooperative system in a

biomedical research environment requires a high de-

gree of ﬂexibility and extensibility. Distributed data is

accessed in different levels of granularity considering

data privacy issues, it is annotated and analysed. We

use the above-mentioned scenarios to deduce general

requirements a CSCW system has to comply with.

• R(1) Flexible Integration and Composition of

Services: A multitude of data processing and

data analysis tools exists in the biomedical con-

text. Some tools act as complementary parts in a

chain of processing steps. For example, to detect

genes correlated with a disease, gene expression

proﬁles are created by measuring and quantifying

gene activities. The resulting gene expression ra-

tios are normalised and candidate genes are pres-

elected. Finally, signiﬁcance analysis is applied

to identify relevant genes (Tusher et al., 2001).

Each functionality may be provided by a sepa-

rate tool (GENESPRING, 2005; Sturn A, 2002).

In some cases tools provide equal functionality

and may be chosen as alternatives. Through ﬂex-

ible integration of tools as services with standard-

ised input and output interfaces a dynamic com-

position of tools may be accomplished. From the

system’s perspective services are technology neu-

tral, loosely coupled, and support location trans-

parency (Papazoglou, 2003). Therewith the ex-

ecution of services is not limited to proprietary

operation systems and any service caller does not

know the internal structure of a service. Further,

services may be physically distributed over de-

partments and institutes, e.g. image scanning and

processing are executed in a speciﬁc laboratory

where the gene expression slides reside.

• R(2) Knowledge Creation and Knowledge Pro-

cessing: Cooperative medical activities fre-

quently comprise the creation of new knowledge.

TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED FLEXIBLE COLLABORATIVE SYSTEMS - A

Basic Framework Applied to Medical Research

233

Data sources are linked with each other, similari-

ties and differences are detected, and involved fac-

tors are identiﬁed. Consider for example a set of

genes which is assumed to be strongly correlated

with the genesis of a speciﬁc cancer subtype. If

the hypothesis is veriﬁed, this information may

be reused in subsequent research. Thus, means

to formalise knowledge, share it in arbitrary con-

texts, and deduce new knowledge are required.

• R(3) User and Role Management: The CSCW

has to be able to cope with the organisational

structure available in the institutes and research

groups of the hospital. Data protection directives

have to ﬁt the system’s access right model. Even

though, the model has to be ﬂexible to allow the

creation of new research teams and information

sharing across organisational borders.

• R(4) Transparency of Physical Storage: Al-

though data may be stored in distributed locations,

data retrieval and data storage depend solely on

access rights, irrespective of the physical location.

Hence, the complexity of data structures is hid-

den from the end user’s perspective. Appropriate

search, addition, and transformation mechanisms

have to be offered by the CSCW system.

3 ARCHITECTURE OVERVIEW

The Wasabi framework enabling collaborative work

as described in the previous sections is a service-

oriented architecture (Schulte et al., 2007). It bases

on the JBoss Application Server (AS) in order to fulﬁl

requirements for enterprise solutions and to provide

the scalability and performance needed for a CSCW

system which is used in a distributed manner like in

the GATiB project. In addition, to fulﬁll the require-

ments previously presented, service orientation is an

important characteristic in the context of ﬂexibility,

adaptability, and maintenance. Note that the service

orientation of Wasabi can be attained due to the un-

derlying JBoss AS also being service oriented.

Service orientation is generally an essential char-

acteristic for a CSCW system since the data stored

might be of arbitrary formats and located in ar-

bitrary repositories. CSCW systems offer a wide

ﬁeld of collaboration activities, supporting its ﬂexible

way/adaptability of storing and handling of data, no

longer limiting its activities to speciﬁc environments

or applications. This Wasabi’s functionality enables

on the one hand to store data in databases or in ﬁle

systems reachable via the network. On the other hand

it can link already existing databases and data sources

to provide users access to those data under consid-

eration of their rights. This is particularly beneﬁcial

for the GATiB project since the data collected in dif-

ferent hospitals and institutes is thereby available for

the whole community without transferring the data for

collaboration to a speciﬁc repository.

In the following Section 3.1 presents the under-

lying idea of organising information sharing and co-

operative work for the collaborative system. After-

wards Section 3.2 describes the key components of

the server architecture.

3.1 Knowledge Spaces

Knowledge spaces are our representation and struc-

ture of presenting the different context types of col-

laboration situations to the users/actors in the collab-

oration process. The organisation of individuals col-

laborating in groups has to be visualised in a useful

way reﬂecting the real world. Users need a knowl-

edge space as a virtual environment for their collab-

oration activities in which they can meet and work.

We deﬁne a knowledge space as an actual use case

in a collaboration context. For example a knowledge

space might be created for an actual clinical study or

a collaborative patient analysis.

Although knowledge spaces are separate virtual

concepts, data and knowledge exchange between

knowledge spaces is encouraged. An important capa-

bility is to upload data or link to remote data which is

already available. To sufﬁce the different users’ needs

for their collaboration, they should organise and struc-

ture the knowledge space individually. Consequently,

users are responsible for the organisation of their

knowledge space which implies highest ﬂexibility.

However, reorganisation of one speciﬁc knowledge

space might be restricted to a limited group of users in

order to avoid unauthorised modiﬁcations. Such self-

organisation conforms to the self-organisation forms

of knowledge in the Web 2.0 (tagging). Further con-

cepts of the Web 2.0 like annotating available infor-

mation is important for a useful CSCW system.

3.2 Server Architecture

The server consists of four main components. First,

the core of the Wasabi enterprise server architec-

ture implements the framework for a CSCW system

and aggregates all services of Wasabi (see Figure 1).

Knowledge spaces are realised by rooms which are

containers for users and documents (any uploaded

content). Furthermore rooms can be connected via

exits. To realise the persistence of all objects of the

server core in a simple way, they all inherit from the

ICEIS 2008 - International Conference on Enterprise Information Systems

234

WasabiObj class. To implement a SOA with this

underlying core classes, the access to objects of the

Wasabi core is realised by the second component of

the Wasabi server - the EJB services (see Figure 2).

For example the UserManager is the responsible ser-

vice for handling, modifying, and extracting informa-

tion of user objects. Their responsibility classify the

EJB services as basis services since they implement

basic functionalities on the objects of the Wasabi core.

Since their tasks focus on the modiﬁcation and pro-

visioning of data stored in the Wasabi core objects,

these can be further classiﬁed as data-centric basic

services. EJB services are also used to realise ﬂexi-

ble user authentication mechanisms as well as to be

adaptable to various content backends/repositories.

Wasabi-Core

Document

WasabiObjLink

Exit

Room Container User

Group

* 1

0..1 1

Communication-

Services

Remote-

API

EJB-

Services

Figure 1: Components of Wasabi Core.

All EJB services handling objects of the Wasabi Core

are realised as service beans, i.e. they implement the

EJB Management Interface. Service beans have the

constraints that only one instance is created on the

server. By using service beans no transaction control

has to be performed since all users or remote/intern

services are interacting with the same instance of the

EJB service. For the internal usage of the EJB ser-

vices local interfaces are implemented which might

offer more or different functionalities than accessible

from remote services or clients.

Third, the remote API provides an interface for

server-client communication (see Figure 3). There-

with a common interface can be used to send requests

to different adapted services, if the services expect the

same input data. This simpliﬁes the enhancement of

Wasabi Beans by adapting new services with little ef-

fort. Note that the remote API might differ from the

local API of the EJB services. The fourth component

is responsible for the message exchange with adapted

services. The server core is enhanced by services real-

ising communication features which are essential in a

cooperative system (see Figure 4). The speciﬁc com-

munication services generate the outgoing messages

according to the deﬁned interfaces of the remote web-

services and processes and extracts information from

incoming messages.

EJB - Services

LDAP

Hibernate

JPA

JCR

User

Manager

Group

Manager

Document

Manager

Object

Manager

Figure 2: EJB Services Realising Service Orientation.

Remote - API

External

Application

External

Application

Service - Consumer

Figure 3: API for User Interaction.

Communication - Services

....

Index / Search

Instant - Messaging

Mail

Figure 4: Core-Enhancement for Communication Support.

TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED FLEXIBLE COLLABORATIVE SYSTEMS - A

Basic Framework Applied to Medical Research

235

4 ARCHITECTURE DETAILS

The example application had shown that data used for

collaboration has to be differentiated. Furthermore,

the environment for a collaboration has to be speciﬁed

and the users have to be represented in the CSCW

system in a suitable way. This section details how

the requirements of the Section 2.0.1 are fulﬁlled in

the architecture.

4.1 R(1): Flexible Integration and

Composition of Services

The requested ﬂexibility is achieved by the service

orientation of the Wasabi-CSCW infrastructure. To

access data from internal objects, appropriate remote

and local services exist. Intermediate services are

used to invoke remote webservices and to handle in-

formation from those. Due to the service orientation,

providing the user a composition of services requires

little effort for internal services as well as a mixture

of internal and remote services (or service composi-

tions).

Arbitrary services can be linked to the CSCW sys-

tem. Intermediate services can use either the local or

remote API of the server’s EJB services. These are re-

sponsible for generating appropriate messages for re-

mote services and to handle incoming messages. The

handling of incoming messages consists of calling an

adequate service and method of the Wasabi server and

to transform the data in the requested format. Hence,

a ﬂexible integration or composition of services is

possible.

4.2 R(2): Knowledge Creation and

Knowledge Processing

A knowledge object is the result of a non-empty se-

quence of functions applied to at least one data ob-

ject. Generally, each knowledge object enhances in-

formation currently available in data objects. We dis-

tinguish various types of knowledge objects: graphi-

cal objects (plots, diagrams, etc.), tabular objects pre-

senting summarised information (e.g. the results ag-

gregation operations), and dependency structures (e.g.

ontologies). Semantic structures may be used to for-

malise knowledge. Relationships between objects or

groups of objects are stored in a processable way in

order to build a knowledge repository. This reposi-

tory integrates conclusions from various knowledge

spaces allowing to explore consolidated knowledge

and deduce new knowledge. Making knowledge per-

sistent and processable allows us to fulﬁl requirement

R(2). The post-processing of accessible data objects

is a speciﬁc but also important criterion for a CSCW

system supporting medical research. Since no limita-

tions concerning the data stored and accessed should

exist, the methods of post-processing should also not

be limited. The CSCW system has to support all re-

quired post-processing steps and must be ﬂexible to

integrate new functionalities with little effort.

The post-processing of data can be performed in

various ways and with different intentions and anal-

ysis targets. The methods of post-processing signif-

icantly depend on the users as well as the underly-

ing data objects. Hence, support a ﬂexible framework

for integrating new post-processing methods and new

knowledge objects, essential for the collaborative sys-

tem. The service oriented architecture supports an

easy linking of services which is more ﬂexible than

integrating speciﬁc functionalities into the collabora-

tive system itself. The linking is performed by ded-

icated services. The Wasabi core has data objects

and knowledge objects to fulﬁl the distinction as re-

quested.

For GATiB the CSCW system should provide dif-

ferent views on the same data, for example for the

staff working directly with a patient have full access

to all her details, whereas researchers only see statis-

tics about multiple patients. Showing statistics im-

plies that a post-processing – evaluating the accessible

data and generating the statistics – is performed auto-

matically by the collaborative system. (Stark et al.,

2008) presents several scenarios for the usage of the

cooperative system in GATiB. The Wasabi framework

interacts therefore with a speciﬁc service which gen-

erates the statistics according to the content requested

by the user. In order to ensure the data privacy, the

data is read out by the Wasabi’s service which trans-

fers it as a SOAP message to the service responsible

for post-processing. The output is then stored by the

Wasabi server as a knowledge object itself before pre-

senting it to the user. The user can then decide to store

the knowledge object which enables her to add anno-

tations. Multiple users can generate different statis-

tics on the same data which results in the generation

of multiple knowledge objects.

4.3 R(3): User and Role Management

An individual is a person participating in collabora-

tive acts. Individuals may be for instance researchers

from the medical or biological domain, medical stu-

dents, or project managers. Individuals are cate-

gorised into internal and external persons in order

to differentiate between access to sensitive patient-

related and research-related data as well as access to

anonymised and summarised data.

ICEIS 2008 - International Conference on Enterprise Information Systems

236

The CSCW system ensures that individuals are dis-

tinguished in order to reﬂect the activities of one in-

dividual in the real world as a one-to-one mapping in

the virtual world. This is important in the context of

collaboration (discussions, annotations, etc.) as well

as for the data access (authentication and authorisa-

tion). The organisational mapping accomplishes the

user and role management speciﬁed in R(3).

The Wasabi-CSCW infrastructure creates a user

object for each CSCW system user, who are distin-

guished internally when checking access rights and

service invocation.

Person A

Person C

Person B

Gruppe 1

Ressource 1

Ressource 2

Ressource 3

Ressource 4

Rolle A

Rolle B

Rolle C

Rolle D

Rolle E

Figure 5: Role Assignment.

Individuals having the same access rights and using

the same set of functions may be integrated to groups

and certain roles may be assigned to users/groups.

Roles are used to adjust the access of individuals

and groups to resources of the CSCW system. As

shown in (see Figure 5), a role can be used to bun-

dle access parameters. The Wasabi-CSCW infrastruc-

ture supports a ﬁne-grained deﬁnition of roles and

groups for users. Users can be assigned to an arbi-

trary number of groups which can also be related to

other groups. Hence, individuals can be part of sev-

eral groups and a group can consist of subgroups. For

example all users working in the same hospital are in

one group, denoted hospital X. The ”regular“ medical

staff is allowed to read and write all data which is pre-

sented patient-centered. The medical students associ-

ated to the hospital are not allowed to see all sensitive

data. Accordingly, the medical staff is a subgroup

of hospital X. The EJB services GroupManager and

UserManager (see Figure 2) are responsible for the

user management. They ensure that logins are unique

within a domain or within the whole system. Further-

more, rights are deﬁned when generating a user and

additional access rights can be set by assigning users

to groups.

Virtual knowledge spaces (rooms) can have links

to other rooms and links to data stored in any linked

repository. A room supports several perspectives

which enables users in the same room to see the data

in a different way and with a different level of details

shown. The perspectives for a user can be deﬁned

by roles. In particular, the various level of details

is a strong requirement in the context of the GATiB

project since the data available is often sensitive and

data privacy has to be supported by the system. In ad-

dition to the perspectives which might vary from user

to user, the Wasabi framework also provides a detailed

authorisation model. For each user the access rights

can be deﬁned, and before granting the user write/read

access, these are checked by a speciﬁc service of

the framework. Since data stored not directly in the

Wasabi server has to be prevented from unauthorised

access, security issues have to be considered not only

in the context of accessing data through the collabora-

tive system’s user interface. Data stored in a databases

is protected from unauthorised access by requesting

user and password during the connection establish-

ment with the database. Data stored in a ﬁlesystem

requires also a similar access validation. This require-

ment was considered in the Wasabi framework by us-

ing an LDAP server. The UserManager is responsible

for the user management within the CSCW system

and consequently it interacts with the LDAP server.

This solution has the advantage that access rights are

checked within the collaborative system and in ad-

dition also in the repository. Furthermore, interact-

ing with an LDAP server has the advantage that of-

ten already established repositories work with LDAP

servers and consequently no additional effort is nec-

essary for transferring the right speciﬁcation into the

Wasabi framework. Note that because of the service

orientation of Wasabi, other authorisation services

providing similar functionalities like LDAP servers

can be integrated.

4.4 R(4): Transparency of Physical

Storage

A ﬂexible and adaptable CSCW system for various

ﬁelds of application must not have limitations con-

cerning the data stored and accessed. Since arbitrary

data types have to be supported, applications used to

visualise or modify the data should be made available

within the CSCW system. Content made available

in the CSCW system is handled as a document ob-

ject which has no limitations concerning data types,

repository, or size. Access rights are checked inter-

nally before sending requests to the repository. We

propose ﬂexible levels of data granularity to satisfy

requirement R(4).

The data used for analyses in the GATiB project

is primarily stored in databases. The relational con-

cept of data objects has to be transferred into an ob-

ject oriented concept presented in virtual knowledge

spaces. Each tuple of information stored in one row

of a database table has to be mapped into an object.

Most developments using the Java Persistence API

TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED FLEXIBLE COLLABORATIVE SYSTEMS - A

Basic Framework Applied to Medical Research

237

start with the deﬁnition of classes from which then

the databases are automatically generated. Since the

data used in the GATiB project is already stored in

different databases basing on various relational mod-

els, the object relational mapping has to be “manu-

ally” performed in the Wasabi framework by an ap-

propriate service. Therefore each data object inher-

its from a WasabiObj class deﬁned in Wasabi (see

Figure 1). Internally core objects like individuals,

groups, and rooms are persistently stored in the cen-

tral database. Since the related classes are subclasses

of WasabiObj, it is necessary to assign to each object

a unique identiﬁer (UUID). Storing all WasabiObj in

a central database, support to generate a unique iden-

tiﬁed for a WasabiObj, i.e. no room or individual can

have the same UUID.

Metadata of document objects is also stored in the

central database by using the persistence capabilities

of entities. In the central database a reference to the

repository is held in the associated ﬁle, data or knowl-

edge object is stored. Through the service orientation

of the Wasabi framework, for each database having

a different relational model a speciﬁc service can be

integrated. Before accessing data of a repository, the

Wasabi’s DocumentManager selects the correct ser-

vice according to conﬁguration options for the repos-

itory to which the related entry in the central database

refer. Afterwards the DocumentManager engaged this

service to perform the object relational mapping. The

usage of a repository-speciﬁc service enables infor-

mation stored in databases to be handled with differ-

ent underlying models in a ﬂexible way. Note that

repositories with the same relational model can use

the same Wasabi service since the connection param-

eters are read out dynamically.

The support of different databases can be realised

through using different services. For cooperative sys-

tems in general a mime-type speciﬁc repository se-

lection is realised in the ﬁrst Wasabi prototype having

several advantages. First of all, speciﬁc content types

might ask for speciﬁc repository features. For exam-

ple a video can be stored on a video server provid-

ing features a “normal” repository does not provide,

like video-streaming. Furthermore, small objects like

annotations can be stored within a database. How-

ever storing text-documents in a database is not very

efﬁcient since databases transform each object to a

speciﬁc data type; accordingly these will be stored

as character- or binary large objects (CLOB/LOB).

The type transformation implies that for writing con-

tent into the database a serialisation is necessary; for

reading content, the objects have to be de-serialised.

This effort is time-intensive for large ﬁles and conse-

quently in this case the usage of ﬁle systems as repos-

itories should be preferred. The conﬁguration which

mime-type should be stored in which repository (any

ﬁle system or database) is speciﬁed within an XML

ﬁle.

When using ﬁle systems two different approaches

for the data organisation can be followed. In one so-

lution, the hierarchy of the knowledge spaces can be

reﬂected in the data organisation in the ﬁlesystem,

e.g. ﬁles are named as within the CSCW system and

the folder organisation conforms to the room name

and hierarchy. Reﬂecting the hierarchy in the ﬁlesys-

tem organisation has the advantage that the ﬁlesystem

could be accessed via WebDAV or FTP from individ-

uals not registered within the CSCW system who are

interested in the content. Of course, such an external

data access has to be protected from unauthorised in-

dividuals in order to ensure data privacy. If the data

stored within the ﬁlesystem should be not available

apart from the CSCW system, the ﬁles can be stored

in an arbitrary hierarchy and the UUIDs assigned to

the WasabiObj can be used as ﬁlenames. This alter-

native solution has the advantage that an additional

security mechanism is integrated. Even if an unau-

thorised individual can see the data structure, they can

only guess what is in the ﬁles, as the ﬁle type is not

reﬂected within the name.

5 RELATED WORK

There exist a strong demand for architectures allow-

ing the developers to encapsulate information tools as

services so that clients can access without knowledge

of or control over their internal workings. These re-

quirements are implemented by service-oriented ar-

chitectures (SOAs) since these are suitable for com-

plex environments with changing demands (Foster,

2005).

Today’s CSCW systems, for instance Moo-

dle (Rice, 2006) or Open-sTeam(Bopp et al., 2006)

are mainly focused on using a single server, for this

reason all collaborative services are provided by one

system. Examples for CSCW systems that involve

ﬂexible service oriented architectures to interact with

other existing applications are rare.

A number of international projects have focussed

on the development of collaborative systems, not only

to support collaborative work, but also to support the

needs of medical research. As an example, based

on web services the myGrid project (Stevens et al.,

2003) in the UK uses bioinformatics as their major

focus. Furthermore, the Japanese BioGrid (Nakamura

et al., 2004) project provides a Grid infrastructure to

fulﬁl the needs of computational power in life sci-

ICEIS 2008 - International Conference on Enterprise Information Systems

238

ence by making available sufﬁcient power to com-

plete their complex analyses. In particular, these Grid

developments differ from our Wasabi infrastructure in

the intention of usage. Wasabi is a framework for

a CSCW system, whereas myGrid and BioGrid are

middlewares for running computational jobs (medi-

cal analyses) and for enabling remote data access in a

Grid infrastructure. Since Wasabi is a SOA and med-

ical analyses of data stored in the CSCW system are

performed by using remote web services, even Grid

services can be invoked from a CSCW system build-

ing on Wasabi. Consequently, services providing the

computational power requested for medical analyses

can be linked. Hence, such extensions can be imple-

mented in Wasabi with little effort because of its ser-

vice orientation.

6 CONCLUSIONS

Since collaboration activities are very similar in var-

ious ﬁelds of application, a CSCW system should

be ﬂexible and adaptable for the usage in differ-

ent scopes. This paper determines the requirements

for such a CSCW system exemplarily in the GATiB

project which has many restrictions because of the us-

age of highly sensitive data. To fulﬁl its requirements

(see Section 2.0.1), the presented Wasabi CSCW in-

frastructure is a service oriented architecture and is

able to handle arbitrary data types and repositories.

The ﬂexible object model of the Wasabi core supports

an adjustment for speciﬁc ﬁelds of application and is

therewith not limited to the usage in GATiB.

Requirements in the context of GATiB are ful-

ﬁlled through the SOA, persistence layers, as well as

the ﬂexible Wasabi core: The user and role manage-

ment is realised through the Wasabi core as well as

the support of cooperative functions, knowledge pre-

sentation, and knowledge processing. The service ori-

entation enables the transparency of physical storage

as well as ﬂexible integration and composition of ser-

vices.

The requirements identiﬁed for GATiB can be

found in arbitrary ﬁelds of application. However, in

most scopes the requirements are less strict. Hence,

the Wasabi CSCW infrastructure applied to GATiB

can be adapted for other ﬁelds of collaborative work.

REFERENCES

Asslaber, M., Abuja, P., Stark, K., Eder, J., Gottweis, H.,

Trauner, M., Samonigg, H., Mischinger, H., Schip-

pinger, W., Berghold, A., Denk, H., and Zatloukal, K.

(2007). The genome austria tissue bank (gatib). In

Pathobiology, volume 74, pages 251–258.

BIOBANK (2007). A biobank for the advancement of

medicine. http://www.bioresource-med.com.

Bopp, T., Hinn, R., and Hampel, T. (2006). A service-

oriented infrastructure for collaborative learning in

virtual knowledge spaces. In Education for the 21st

Century, pages 35–44.

Foster, I. (2005). Service-oriented science. Science,

308(5723):814–817.

GEN-AU (2007). Genome research in austria.

http://www.gen-au.at/english/content.jsp.

GENESPRING (2005). Genespring, cutting-edge tools for

expression analysis. http://www.silicongenetics.com.

Nakamura, H., Date, S., Matsuda, H., and Shimojo, S.

(2004). A challenge towards next-generation research

infrastructure for advanced life science. New Gen.

Comput., 22(2):157–166.

P. Samarati, L. S. (1998). Protecting privacy when dis-

closing information: k-anonymity and its enforcement

through generalization and suppression. In Proceed-

ings of the IEEE Symposium on Research in Security

and Privacy.

Papazoglou, M. P. (2003). Service-oriented computing:

Concepts, characteristics and directions. In Web Infor-

mation Systems Engineering (WISE 2003), volume 00,

page 3, Los Alamitos, CA, USA. IEEE Computer So-

ciety.

Rice, W. (2006). Moodle E-Learning Course Development.

Packt Publishing.

Schulte, J., Hampel, T., Bopp, T., and Hinn, R. (2007).

Wasabi framework – an open service infrastructure for

collaborative work. skg, 0:242–247.

Stark, K., Eder, J., and Zatloukal, K. (2006). Priority-

based k-anonymity accomplished by weighted gener-

alisation structures. In DaWaK 2006: Proc. of the

8th International Conference on Data Warehousing

and Knowledge Discovery, Lecture Notes in Com-

puter Science, Volume 4081. Springer Verlag.

Stark, K., Schulte, J., Hampel, T., Schikuta, E., Zatloukal,

K., and Eder, J. (2008). Gatib-cscw, medical research

supported by a service-oriented collaborative system.

In The 20th International Conference on Advanced In-

formation Systems Engineering (CAiSE’08).

Stevens, R. D., Robinson, A. J., and Goble, C. A. (2003).

mygrid: personalised bioinformatics on the informa-

tion grid. Bioinformatics, 19 Suppl 1.

Sturn A, Quackenbush J, T. Z. (2002). Genesis: Cluster

analysis of microarray data. In BIOINFORMATICS

-OXFORD-, volume 18(1), pages 207–208.

Tusher, V., Tibshirani, R., and Chu, G. (2001). Signiﬁ-

cance analysis of microarrays applied to the ionizing

radiation response. In Proceedings of the National

Academy of Science, volume 98, pages 5116–5121.

TOWARDS THE NEXT GENERATION OF SERVICE-ORIENTED FLEXIBLE COLLABORATIVE SYSTEMS - A

Basic Framework Applied to Medical Research

239