Deliverable No. 51

walkingceilInternet and Web Development

Oct 22, 2013 (3 years and 9 months ago)

173 views



Programme

Integrating and Strengthening the European Research

Strategic Objective

Networked business and government

Programme Title

The Virtual Research Lab for a Knowledge Community in Production

Network of Excellence

Acronym

VRL
-
KCiP


Project No

NMP2
-
CT
-
2004
-
507487

VRL
-
KCiP


Work package

Knowledge Management

VRL
-
KCiP
-

WP No.

JRA


WP3

Deliverable No. 51


Newly developed knowledge and communication

management methods


Leading Organisation: MTA SZTAKI


Contributing Tasks: T26

T26:
Definition of knowledge management methodologies and tools for sharing knowledge and applications for
demonstration inside the network

30 May
, 2005

Version 1.0

Dissemination level: CO

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


2
/
56

Versioning and contribution history

Version No.

Name

Organisation

D
escription

Date

1.0


MTA SZTAKI


30 May, 2005







Authors

Name

Organisation

Email

József Váncza

MTA SZTAKI

vancza@sztaki.hu

Zsolt Kemény

MTA SZTAKI

kemeny@sztaki.hu

P
éter Egri

MTA SZTAKI

egri
@sztaki.hu

Zsolt J
ános Viharos

MTA SZTAKI

viharos
@sztaki
.hu

Tetsuo Tomiyama

TU Delft

t.tomiyama@wbmt.tudelft.nl

Matt Giess

BATH

enpmdg@bath.ac.uk

Myriam Lewkowicz

UTT

myriam.lewkowicz@utt.fr

Nada Matta

UTT

nada.matta
@utt.fr

Gunilla
Sivard

KTH

Gunilla.Sivard@iip.kth.se

Ivan Vengust

KOGAST

ivan.vengust@fs.u
ni
-
lj.si


Referenced documents

Document (File)

Comment

JRAWP3_
D50
.doc

Deliverable D50 “State of the art knowledge and
communication management techniques available at the
partners”

KCIP_JRAWP3_Questionnaire_Ans
wered.xls

Compiled results of the
JRA
-
WP3

Q
uestionnaire


The following short names are used when referring to the organisations participating in VRL
-
KCiP:

No

Participant name

Country

Town

Short name

1

Caisse des dépôts et consignations

F

Paris

CDC

2

Institut National Polytechnique de Gren
oble

F

Paris

INPG

3

University of Twente

NL

Enschede

UT CIPV+

4

Franhaufer IPK

G

Berlin

FhG/IPK

5

ITIA CNR

I

Milano

ITIA

6

University of Bath

UK

Bath

Bath

7

Fundation TEKNIKER

S

Eibar

TEKNIKER

8

University of Patras

GR

Patras

UPAT
RAS

9

Kungliga Tekniska Höaskolan

S

Stockholm

KTH

10

Hungarian Academy of Sciences

HU

Budapest

MTA SZTAKI

11

University of Lubljana

SL

Lubljana

KOGAST

12

Universitaet of Stuttgart

G

Stuttgart

USTUTT

13

Israel Institute of Technologie

IL

Haifa

TECHNION

14

Ecole centrale de Nantes

F

Nantes

IRCCyN

15

Université de Technologie de Troyes

F

Troyes

UTT

16

Universit. Politechnica of Timisoara

RO

Timisoara

IEL
-
UPT

17

Ecole Polytechnique Fédérale de Lausanne

CH

Lausanne

EPFL

18

Un
iversity of Durham

UK

Durham

UoD

19

Delft University of Technology

NL

Delft

TU Delft

20

Eindhoven University of Technology

NL

Eindhoven

ECIS TUE

21

Politechnica Poznanska

PL

Poznan

PUT
-
VIDA

22

Pôle Productique Rhône Alpes

F

St Etienne

PPR
A

23

University of Stellenbosch

SA

Stellenbosch

GCC, US

24

Politecnico di Milano

I

Milano

Poli
-
Milano

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


3
/
56

Table of Contents


1

Introduction

................................
................................
................................
..............................
4

2

Our approach for collecting competencies

................................
................................
...............
7

2.1

JRA
-
WP3 Questionnaire

................................
................................
................................
...
8

2.2

Building initial Conceptual Maps in s
pecific technical areas

................................
..........
9

3

Summary of answers given to the questionnaire


a knowledge management perspective

...

10

4

Text mining
for capturing competence

................................
................................
..................

12

4.1

Text mining

................................
................................
................................
....................

12

4.2

Related work

................................
................................
................................
...................

13

4.3

Short, fixed
-
form essays

................................
................................
................................
..

14

4.3.1

The template

................................
................................
................................
................................
...

14

4.4

Examples for short essays

................................
................................
..............................

15

4.5

Comparative evaluation of text miners

................................
................................
...........

18

5

Requirements towards the Knowledge Management system of the VRL KCiP network

......

23

5.1

Position

................................
................................
................................
...........................

23

5.2

System architecture

................................
................................
................................
.........

24

5.3

Typical user processes of the Knowledge Management Syste
m

................................
....

24

5.3.1

Use Case


Creating and updating an individual profile

................................
.........................

25

5.3.2

Use Case
-

Creating and updating an institutional pro
file

................................
......................

26

5.3.3

Use Case


Querying
the

system

................................
................................
................................
.

26

5.3.4

Use Case


Browsing

the conceptual map

................................
................................
................

27

5.3.5

Use Case


Visualizing

an item/essay

................................
................................
........................

27

5.4

A
data
model for the Knowledge Management System

................................
.................

27

5.4.1

Keeping track of researchers and institutions

................................
................................
...........

28

5.4.2

Mapping of concepts

................................
................................
................................
....................

28

5.4.3

Mapping documents, including ess
ays

................................
................................
.......................

30

5.4.4

Connecting the three expertise groups

................................
................................
......................

30

5.5

Supporting cooperative work

................................
................................
..........................

32

6

Conclusions

................................
................................
................................
.............................

33

7

References

................................
................................
................................
...............................

34

Appendix
: further examples for fixed
-
form essays

................................
................................
..........

36

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


4
/
56


1

Introduction

The central topic of JRA
-
WP3 is Knowledge Management that includes approaches allowing to
handle, share, and reuse k
nowledge in an organization.
Task 26, in particular, is aimed at giving
definitions of knowledge management meth
odologies and tools for sharing knowledge. In
another deliverable, D50, we have summarized the results of a network
-
wide survey and have
given an account of the state
-
of
-
the
-
art knowledge and communication methods available at the
partners. Now, in this do
cument, we propose novel knowledge management methods tailored to
the needs of the VRL KCiP.

We depart from the fact that members of the VRL KCiP community form a loosely connected
network of various organizations and people, each having competencies and
disposing over some
information resources in distinct, mostly production related domains. The competence of the
whole network covers almost all production domains. The network is, however, large,
competencies of people are partly implicit and most of the i
nformation resources are hidden.
Hence, there is a need to establish a novel, conceptual network


a kind of
competence

map

--

on
top of the existing network that supports one in



navigating through the VRL KCiP NoE,



finding the right persons and organizat
ions, as well as in



locating the right resources (documents).

Such a map would provide

a match between people (both individual and in organization), their
domain of interest and the knowledge
(re)
sources they posses and/
or contributed to in any way.
It ca
n be considered an
appropriate "yellow pages" of our community.

Such a competence map
that is based on shared and common understanding of our domains of interest and our
competencies is essential in search, information exchange, and discovery. Hence, it is

a
prerequisite for operating the KCiP NoE as a Virtual Research Laboratory.

Our
goal
in JRA
-
WP3, and in particular in Task 26, is to define a process that helps the KCiP
community to build on top of the network of its resources (typically documents) a ne
twork of
related concepts. We consider this a
knowledge management problem

that calls for appropriate
computer assistance. Hence, we are going also to analyze the requirements towards the services
of a knowledge management system that helps KCiP members to

build up and use this
WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


5
/
56

competence map, together with the underlying network of resources, organizations and people.
Note that this specification is purely functional and does not concern implementation issues.

We can distinguish two
main functions

of
a
kn
owledge management tool suppor
ting the

VRL:



to
characterize and
identify people

as well as institutions of

the network

and to make
their knowledge resources available,



to support cooperative work of network members in a virtual environment.

In this docume
nt, we focus on the first kind.

Note that s
everal types of tec
hniques can be used
for capturing and organizing knowledge encoded in documents (cf. deli
v
e
rable
D
39)
.

In what
follows, we focus on taxonomy and ontology
-
based
representation
methods

because the
y are
essential for drawing
a competence map based on concepts
relevant in the professional interest of
the KCiP members.

There are various structures for organizing knowledge encoded in documents. Below, we make a
short distinction between them.



Taxonomy

is a

system of articulated concepts
, in a typically one dimensional classification
scheme. A so
-
called

faceted taxonomy
describe
s

multiple
classification schemes, each of
which describes the properties or characteristics

of a particular subject domain
.



Te
rminology

is a kind of taxonomy together with relationships among the terms

(not
necessarily hierarchical) represented
as
axioms
. It includes also
theorems

derived from the
axioms. Typical relations are
super
-
sub, ownership, be
-
a
-
part
-
of, equivalent, etc.



Ontology

can be considered a terminology together with
domain dependent knowledge (for
example, usually a shaft has a cylindrical shape)
. Specifically, it is an e
xplicit specification
of the conceptual structure of a given domain that is integral part of t
he consensual
knowledge of a community.

It is usually expressed in a logic
-
based language that makes
possible to distinguish classes, instances, properties, relation and functions in a clear
-
cut,
consistent way. Consensual means that the whole community ha
s a common
understanding both on the content and form of the expressed knowledge. Finally, an
ontology can facilitate also machine processing, automated reasoning and inter
-
operability
of different computer systems.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


6
/
56



Concept maps (conceptual maps, topic map
s)

are knowledge representation methods that reveal
the concepts of a domain and their underlying relationships using a map metaphor.
Concept maps have a direct mapping to and from natural language representations and a
graph
-
based graphic notion designed
for human readability.


This document is structured as follows: First, we present the JRA
-
WP3 action plan that was
aimed at capturing the competencies of KCiP partners in three technical domains (design and
virtual prototyping, manufacturing processes, pr
ocess simulation). A key element of this plan was
to make a questionnaire
-
based survey. After having discussed lessons of this work we suggest a
new approach for constructing a competence map of the KCiP community. Key element of the
method is the applicat
ion of
text mining
techniques. Hence, we give an overview of available text
miner systems and propose an outline for the texts to be processed this way. General
requirements towards the services of the knowledge management system of our network are
present
ed in form of use cases. Finally, we give an initial specification for the data model of the
knowledge
-
related information to be stored in such a system.


WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


7
/
56

2


Our approach for collecting competencies

In order to achieve the

goal
s of
JRA
-
WP3
,
partners set up
an action plan, together with

an
appropriate schedule for performing Tasks T23
-
27 and producing deliverables D38
-
52 (see figure
below).


Figure 1. Action plan of
JRA
-
WP3


The action plan contains the description of actions, participants, internal due date
s as well as the
responsible persons. After approval of the participants, further work of JRA
-
WP3 has been
organized according to this action plan.

In the action plan we determined to collect competencies

of the
VRL
KCiP partners in a four
-
staged process:

1.

Collecting descriptions of competence together with references to resources from the
whole community.

2.

Based on the responses, building initi
al conceptual maps i
n some selected technical areas
that represent the main classification categories and their rel
ations in a semi
-
formal,
graphical way.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


8
/
56

3.

Based on the responses, requiring detailed contributions from partners who indicated
special experience


hopefully even expertise


in workflow, CSCW and/or knowledge
management, or in the use/development of ontolog
ies.

4.

Refining the conceptual maps towards more expressive and formal representations, like
multi
-
view classification schemes, or, if possible, towards ontologies.

Progress of the action plan was assessed at the
JRA
-
WP3

Workshop in Budapest, 18
-
19, April,
2
005. The Workshop gave also recommendations on implementing conceptual maps in the
selected fields as well as accepted a proposal for acquiring and systemizing network
-
wide, detailed
information about the (individual) members’ competencies and expertise w
ithin a
text mining
project
. This propo
sal is elaborated in Section 4

of this document.

2.1

JRA
-
WP3

Questionnaire

Partners w
ith key responsibilities in
JRA
-
WP3

--

MTA SZTAKI
(
Budapest
)
, IRCCyN (Nantes),
UPATRAS

and
PUT
-
VIDA (Poznan)


developed a web
-
based ques
tionnaire for collecting
information about the competencies, relations and knowledge management practice of the VRL
KCiP members

(for details, see deliverable D50)
.
On the one hand we asked for general
information about the partners’ scientific activities
and results, on the other hand
focused our
interest to the domains of design, production processes, simulation and knowledge management
processes.

Hence, the questionnaire contained the following
sections:

A) Personal Data

B)
Organization's Data

C) Scienti
fic Activities and Results

D) Design and Virtual Prototyping

E) Simulation

F) Production Technologies and Processes

G) Knowledge Management

H) Comments


In the questionnaire there were entries

both
for
structured and unstructured information. For
instance,

the screen below mirrors this twofold feature:

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


9
/
56


Figure
1
. Sample page of the
JRA
-
WP3

Questionnaire


2.2

Building initial Conceptual Map
s

in specific technical areas

Departing from the results of the initial

survey process, Tasks 23, 24 and 25 started to create
conceptual maps in selected technical domains such as:



d
esign and virtual prototyping
,



s
imulation
, and



m
anufacturing processes
.


The deliverables
D41, D46 and D39

sum

up the results of this process, re
spectively,
and present
the first versions of the concep
tual maps
.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


10
/
56

3

Summary

of answers given to the questionnaire


a knowledge
management perspective

The lessons of our extensive survey within the
VRL
KCiP network about the network members’
competencies
and first
-
hand experience in knowledge management can be summed up as follows

(for more details, see D50)
:



Both prior expertise of partners involved in knowledge management (KM) research and
the analysis of answers led to the conclusion that the network ne
eds the services of a KM
system.



The system should support both individual and institutional
VRL
KCiP members in

(1)
identifying their competence
profile, in (2) navigating over the
competence map

of the
network, and (3)

in
work
ing cooperatively with other

network members.



As for the services of the system, members put most emphasis on
retrieval



both keyword
search and category browsing.



There was a definite demand for storing informa
tion in a well
-
structured organization
, for
controlled
content quality
and for regular content updating.



Network members highly estimated the importance both of distributed, self
-
organized
and top
-
do
wn,
hierarchical organization structures (though seemed to prefer the first
one). Finding a trade
-
off between these conflicting

requirements will be a key issue for
the designers of the common knowledge management systems of the KCiP network.



A big majority of network members referred solely to their publications as primary
knowledge sources.



The length and style of contributions
required in free text format varied extensively.



The personal effort
s

of those who contributed varied extensively (note that the
questionnaire had many optional sections and free
-
text entry fields).



Because of the uneven quality and amount of answers comin
g from various partners, the
actual roles and relations of member institutes could hardly be
established without
distortions.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


11
/
56



All in all, neither the web
-
based, nor the subsequent
directed, personal

acquisition
processes could produce information in suffic
ient quality and amount for defining the
competence map

of the network.

At the start of our work it was already clear that a
match between people, their domain of interest
and the knowledge
(re)
sources they posses
can be drawn only in an iterative refineme
nt process.
Now, t
he above points suggest also that

we have to continue the competence acquisition process,
but hint also how to do that. Certainly, further steps must be driven by the definite goal of
designing, implementing, running and maintaining the k
nowledge management system of our
network. In this
acquisition
process:



W
e have to give freedom for both individual and institutional network members for
building up their profile.



At the same time,
these profiles should be constructed according to well
-
de
fined, rigorous
guidelines.



P
rocessing information coming from so many different sources
calls for the application
of at
least semi
-
automatic methods.

After analyzing the responses to the
JRA
-
WP3

questionnaire it became
also
evident that a
directed surve
y process (like our one based on a questionnaire) is too restrictive. Those who are
willing to contribute must have a greater freedom to express themselves, to describe their
domains of interests and competencies, as well as their relations to other partne
rs. Consequently,
we have to work with free text documents. On the other hand, free text documents written
according to different standards (scientific paper, project report
, interim report, web
-
page, etc.
),
in various styles, in longer or shorter lengths
convey information (and knowledge, in the
background)

of very different qualities

and value
s
.

Hence, at the
JRA
-
WP3

Workshop in Budapest, 18
-
19 April, 2005, we decided to make a trade
-
off and suggested the application of
text mining methods

over pre
-
define
d,
fixed
-
form short
essays.
T
his
idea

is
presented in Section 4

below.



WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


12
/
56

4

Text mining

for c
apturing

competence

To start with, we

make the
following
hypothe
sis:
in order to build a knowledge management
s
ystem by the use of a document m
anagement
s
ystem, we ar
e going to describe the knowledge in
documents
.

Note that this assumption concurs well with our findings drawn from the results of the
JRA
-
WP3

survey.

Here we suggest a
method

for collecting,
codify
ing and making accessible competence available
within the
VRL KCiP

network.
The main objectives are as follows
:

1.

Collect
descriptions of domain knowledge

available at the
VRL
KCiP partners in a unified
format.

2.

Set up basic material for supporting information search and retrieval over the above and
all other docume
nts

available at the network members.

The key idea is to call
VRL
KCiP network members for submitting
short,
fixed
-
form essays

of their
domains of interests, results and expertise, and to process, analyze and visualize this corpus by
text mining methods.

4.1

Text mining

Text mining
, also known as
intelligent text analysis
,
text data mining

or k
nowledge
-
discovery in text

(KDT),
refers generally to the process of extracting interesting and non
-
trivial information and
knowledge from unstructu
red text. Text mining

is an

interdisciplinary field which draws on
information retrieval, data mining, machine learning, statistics and computational linguis
tics. As
most information (cc. 70
-
80%) is st
o
red as text, text mining is widely
considered a knowledge
management method

of primary importance
.

There are several ways for applying text mining over the body of short, fixed
-
form
essay
s:



Given pre
-
determined classification schemes


like the conceptual maps developed on the
basis of survey data, or the product and product life
-
cycle related taxonomy developed by
IA
-
WP1


text mining can arrange documents in these classification schemes according
to the content of the
essay
s. The classification can be performed in various views
focusing on e.g., problem domains, applied methods,

people, time, status of knowledge,
etc.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


13
/
56



Given the body of
essay
s with pre
-
determined category labels, text mining methods can
build a model that automatically assigns the right content categories to new, unprocessed
documents. Hence, the short
essay
s can

form the
training set
in a supervised learning
process.



Departing from
tabula rasa
, text mining can arrange documents into clusters according to
their textual similarity. Different similarity measures lead to different categories of the
essay
s. Here the s
hort
essay
s are the subject of an unsupervised learning process. Further
on, text mining can also help in extracting keywords from the documents (see feature
selection).

Of course, the

co
rpus of essays can help not only in building up the competence map o
f the VRL
KCiP, but can provide material for a


preferably electronic


book over the network.

4.2

Related wo
r
k

Though the knowledge management method suggested above is new, the application of text
mining over bodies of relatively short texts is not without
preliminaries. Technological
intelligence has recently applied text mining for empirical
technology mapping



for drawing so
-
called
technology maps that convey emphases, main players and patterns in the development of a
particular target technology [Zhua a
nd Porter, 2002]. The subjects of such investigations are
abstract and patent databases. In particular,
patent mapping

is a target area of text mining because in
contrast to traditional, black
-
box linguistic analysis tools they offer controllable, glass
-
b
ox
methods for patent information professionals [Fattori et al., 2003]. Still over bodies of patents,
text mining techniques have been applied in setting up high
-
technology trends and so
-
called
patent
networks

[Yoon and Park, 2004]. In a wider context, clo
se to our goals in the VRL KCiP, text
mining has been considered a future technology for drawing
roadmaps

for so
-
called disruptive
technologies [Kostoff et al., 2004].
There are several attempts for analyzing the content and
building up the conceptual stru
cture of web
-
based corpora (like short articles, news, abstracts) by
text mining methods [Scharl and Bauer, 2004; Yang and Lee, 2004; Ong et al., 2005]. In the
pharmaceutical and biotech industry, where
the
development of new drugs
needs the review of a
va
st amount

of literature, text m
ining methods are promising candidates for processing and
organizing literature resources [Hale, 2005].

We can
also
learn from information and knowledge management in social sciences and
humanities, where inter
-
disciplinary a
nd inter
-
institutional collaborations


just like in the case of
WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


14
/
56

the VRL KCiP network


are more and more common. In the support of such activities
advanced text mining methods are going to have a growing importance [Bajcsy and McGeer,
2004].

A somewhat si
milar experiment was carried out recently, when a body of texts containing the
descriptions of
1700 E
U

IST projects


each consisting of a few hundred of words


were
analyzed by clustering methods. For further details, see the project
SolEUNet: Data Minin
g and
Decision Support for Business Competitiveness: A European Virtual Enterprise
,
http://soleunet.ijs.si/website/html/euproject.html
.

An example o
f a site with a similar purpose is

Resul
t Center™, a meeting place for Swedish
research in product realization,
http://www.kunskapsformedlingen.se/?lang=en
.

4.3

Short, fixed
-
form
essay
s

The
essay
s will be
read by a large variety of
people,

with different background.
Keeping this in
mind the
essay
s’
style

should be descriptive (emphasize more the 'what' than
the 'how')
wit
hout
too much technical detail. References to illustrations, further readings, links to web

pages are
important, but shou
ld be separated fr
om the main text and put into

extra section
s
.

Length

of the essay should be kept
at a maximum
around
700
-
800 word
s. Each individual is
expected to submit no more than 5 distinct
essay
s

over different topics
. The essays should be
submitted

preferably in ASCII text
, Latex

or MS Word

format.

For collecting and
codify
ing expert

knowledge in a way t
hat is accessible to most of the people,
we will need a clear, well
-
structured description

of what the subject matter is about
. Hence, we
have to d
evelop a
template

for the
essay
s and provide
guidelines

for writing them.

4.3.1

The template

As for the template of fixed
-
form, short essays, we suggest to
adapt the widely accepted template
of short scientific
articles
.

Accordingly, t
he short
essay

template is
structured as follows:



T
itle



the
concept we would like to explain

or piece of work we want to present.



Author(s)

without detailed affiliation data.



Lead
, a kind of teaser
with a few words

about the topic
. The lead
is intended to raise
interest, so it s
hould be k
ep
t short. It should
briefly summarize the most important points
WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


15
/
56

covered in an
essay

in such a way that it could stand on its own as a concise

version of
the
essay
.



Body

with some details of

what the concept is
. One has to t
ry to include related
concepts, links, keywords in the body. Important features are: definition, historical
development, major ap
plications, etc. If the essay is about a project, one can put emphasis
on factors for success and factors to avoid.



Link(s)

to
illustration
s

(photos,

graphics)
,

references,
applications mentioned in the
essay
, people working on the project, etc.



Related c
oncepts
--

super co
ncepts or important subconcepts,

typically those that can
be found in the conceptual maps of the technical areas or in the taxonom
ic scheme
developed by
IA
-
WP
1. F
or example, if the entr
y is 3D solid modeling,
concepts
like

boundary representation, volume representation, constructive solid geometry, etc. are
“related concepts”
.


These could be candidates for
further
entries.




Keywords

selected, preferably, from a pre
-
defined list (like the CIRP Keyword List).



Illustrations
referred to
in the Body or
under
the
Links. Optional.

4.4

E
xample
s for short essays

Consider the following
two
example
s

for a short es
say. Other examples are presented
in the
Appendix of this document.


T
itle

Computer
-
assisted collective problem
-
solving

Lead

We aim at assisting collective problem solving processes which appear for in
stance during a
project, where different types of actors have to work together in a flexible and reactive way. To
assist this co
-
operative design process, it is necessary to answer new types of needs: lack of
coordination formalization, geographical deloca
lization of teams’ members, and difficulty of
building a shared reference frame.


To answer these needs, we propose the Memo
-
net groupware which aim is to provide a
framework for arguments exchanges during design meetings, in a synchronous or
asynchronous

way.

Author

Organization

UTT

Contributing
members

Myriam Lewkowicz

Date

May 2005

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


16
/
56

Body

In Memo
-
net groupware, team members submit some propositions to solve a problem. These
propositions are classified chronologically (see screen priting below), or by author’s name, role
or department. To close a debate, participants could vote f
or a proposition and explain their
position. All the submitted propositions to solve a problem in the team are then memorized with
arguments and constraints which permuted to make a decision.


Use examples:



Discussion management before and after important

meetings: Memo
-
net allows
brainstorming before the meeting, decision making during the meeting, and tracking
new subje
cts to debate after the meeting.



Design Rationale: Memo
-
net allows memorizing decision making process, supports
diagnosis and helps to fi
nd the best repair solution.

Links

Lewkowicz, M., Zacklad, M. 2002, A structured groupware for a collective decision
-
making aid,
EJOR Feature Issue Human Centered Processes, Vol. 136, N2, January 16, 2002, pp. 333
-
339
.

Related

concepts

p
roblem solving, c
ollaboration
, design teams

Key
words

g
roupware,
c
ollective problem solving,
cooperative k
nowledge
m
anagement,
d
e
sign rationale,
decision making


Illustrations



Table
1
. A sample essay about computer assis
ted cooperative problem solving


WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


17
/
56


T
itle

Concurrent design of a
virtual prototype of a h
igh
-
tech product,

and of embedded
software of a h
igh
-
tech product

Lead

Non
-
engineers do not spend time and effort discussing details of a new product behavior and
they do not elaborate on UML models. But non
-
engineers do accept visual simulators well.
All
parties, involved in a projec
t (designers, managers, sales
-

force, potential customers), have to
unambiguously agree on all aspects of a new product at a close
-
up of the definition phase, and
a design phase needs to be short.
Key issues of this work are:



Different surveys quote low su
ccess rate of embedded software projects.



A majority of embedded software projects fail to come in on time.



New products are not defined at the end of Product definition phase.

Body

Development of new systematic approaches to a High
-
tech product design.
T
hese approaches
are aimed at shortening product design time, lowering development costs, and lowering the
amount of entropy in communication within the development process.
In the proposal, two
things are different from common approaches:





a common ground

in discussions on product definition is defined in a form of a virtual
prototype and



we propose such an internal structure of the virtual prototype that permits re
-
use of
most of prototype’s internals further on in the actual product development.


The ne
w concept of simultaneous creation of virtual prototypes and embedded software is
targeted to an ambitious goal of software re
-
use. Internals of a virtual prototype that simulates
product's behavior, and runs on a workstation are re
-
used to govern the actu
al product.

Virtual
prototyping is gaining importance since it is getting feasible. Coding in parallel with modelling is
a tempting approach since a) modelling gives insight into the planned functionality, and b)
concurrent coding saves time for later dev
elopment phases where time is at premise. Till now,
we did two embedded system designs by the proposed approach. It turned out that in both
cases we were able to port close to eighty percent of the virtual prototype code into the final
product. Now, we wil
l continue and share experience with interested collaboration members
about the formalization of a High
-
tech product design concept that has already proved in the
evaluation phase (two successful industrial designs) as sound and feasible.

Typical applicati
ons are realization of high
-
tech products with built
-
in microcontrollers.


Realisable
b
enefits
:



By the proposed method is more than half of embedded software designed and coded
in the Product definition phase. It is a Product design phase where the time is

usually
lost and the time loss needs to be somehow compensated.



The interface between the Product definition and Product design phase is a clean cut,
based on the properties of a virtual prototype. This helps enormously with the costs and
time of a High
-
t
ech product development.



Embedded software, designed and coded by the proposed approach, is platform
portable and it is reusable.

Author

Organization

Uni.
Ljubljana,
Laboratory LAKOS

Contributing
members

Jenko Marjan, Lap Janez,
Butala Peter

Date

May 2005

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


18
/
56

Li
nks

Je
nko M., Medjeral N., Butala P.,
Component
-
based software as a framework for concurrent
design of programs and platfo
rms
-

an industrial kitchen appliance embedded system.
Microprocessors and. microsystems
, 2001, vol. 25, no. 6, str. 287
-
296
.

Jenko M.,
Concurrent development of a visual simulator and of software as a new concept in
product definition, and in embedded sof
tware design.
Proceedings of the fifth international
symposium on tools and methods of competitive engineering,

(TMCE 2004), april 13
-
17,
p.
1
-
11,
2004, Lausanne, Switzerland
.

Related

concepts

virtual prototyping
, concurrent design

Key
words

rapid p
rotot
yping, concurrent modelling
, software componen
ts, mechatronic system design,
h
igh
-
tech product development, software re
-
use,


Table
2
. Example sheet about
c
oncur
rent design of a virtual prototype


4.5

Comparative evaluation of text mi
ners

UTT
made a
survey

research of
commercial text mining
systems
that are able to do

linguistic
analysis of documents in order

to identify and classify main terms in a text.
Such tools were
compared

that can manipulate several languages and can be applied

in several domains.
We
present in the following a comparison of
these

tools under three types of criteria:



Commercial: price, documentation, service, etc.



Technical: architecture, volumetric, etc.



Functional: possible functions, user
-
friendliness, explo
itation of results, etc.

The study included the following systems:

1.

TEMIS

Insight Discoverer
: an E
uropean leader in TextMining.

2.

SAS Text Miner

: used in industry for instance to index EDF (French electric company)
documents.

3.

SPAD/CRM

: developed by resea
rch laboratories, it allows a fine documents exploration
and classifications of big volume of data. .

4.

TROPES ZOOM

: based on Artificial language techniques (Discourse Propositional
analysis and discourse cognitive analysis) , it allows to solve lexical an
d semantic
ambiguity on the language.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


19
/
56

5.

TKS

: proposed by IBM, it allows to index as well as textual and technical data. It can be
used for several domains.

6.

WordMapper

:
a miner that
uses statistic analysis and define a map of terms.

7.

NeuroNav
: a complete s
tatistic environment of text analysis, it offers navigation tools
and classification.

S
ome of these tools have been tested using a paper on mechanical design
[
Smith et al.
].
The
following table
s show

the results of t
his comparison using financial
and
tec
hnical criteria as

c
ompany, cost, data access, linguistics tools used,
automation degree, technical

quality and
diversity, classification methods
, result types and
analysis report. Some notation are used as (*) no
satisfactory to (***) more satisfactory.

WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


20
/
56


Criteria

TEMIS Insight
Discoverer

SAS Text Miner

SPAD/CRM

Tropes/Zoom

TKS

WordMapper

NeuroNav

Company








Origin

France

USA

France

France

Europe

France

France

Name

TEMIS

SAS

DECISIA

ACETIC

IBM/ECAM

Grimmersoft

DIATOPIE

Web Site

www.temis.com

www.sas.com

www.decisia.fr

www.acetic.fr

www.ibm.com

www.grimmersoft.com

www.diatopie.com

Product


Financial aspects








Licence cost

70

000


According to the
request

According to the
request

5 382 €

30

000 €

450 €

According to the
request

Formation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Product


Technical aspects








Server/Client Mode

PC

Server/clie
nt

PC


Server/client

PC


Server/client

PC


Server/client

PC


Server/client

PC


Server/
client

PC


Server/client

Exploitation system



Windows

:


95

98
NT

2000
XP
2003 server

L
inux

Windows

:


95

98
NT

2000
XP
2003 server

Linux

Windows

:


95

98
NT

2000
XP
2003 server

Linux

Windows

:


95

98
NT

2000
XP
2003
server

Linux

Windows

:


95

98
NT

2000
XP
2003 server

Linux

Windows

:


95

98
NT

2000
XP
2003 server

Linux

Windows

:


95

98
NT

2000
XP

2003 server

Linux

Memory Space

Disk Space

128 MB


40 MB client / 300
MB Server

128 MB

200 MB




512 MB

Needed Tools




None

Others

:

Temis pack

None

Others

:

SAS Enterprise Miner

None

Others

:

Excel

None

Others

:

None

Others

:

None

Others

:

None

Others

:

Method Documentation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Save parameters / results

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Ergonomic appreciation

Learning facility

User
-
friendliness


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***

Data a
ccess








Reaching remotely documents

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Need a data extraction tool

yes
no

yes
no

yes
no

y
es
no

yes
no

yes
no

yes
no

Document format




text
ASCII

PDF
Html

MS word

Other: 50

text
ASCII

PDF
Html

MS word

Other

: Lotus

text
ASCII

PDF
Html

MS word

Other

text
ASCII

PDF
Html

MS word

Other

text

ASCII

PDF
Html

MS word

Other

text
ASCII

PDF
Html

MS word

Ot
her

: XML

text
ASCII

PDF
Html

MS word

Other

: ppt,xls

Direct access to Data Base management
system

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no


WP

JRA
-
WP3




Contributing tasks

T26

Ve
rsion

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005




Deliverable No 51.


21
/
56


Cri
t
eri
a

TEMIS Insight
Discoverer

SAS Text Miner

SPAD/CRM

Tropes/Zoom

TKS

WordMapper

NeuroNav

Pre
-
treatement and textual
identification








Document normlalisation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Language




French


E
nglish


Others

: 14


French


English


Others

: 5


French


English


Others

:


French


English


Others

:


French


English


Others

:


French


English


Others

: 1


French


English


Others

: 4

Automatic recognition

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Grammatical recognition

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Nominal groups recognition

yes
no

yes
no

y
es
no

yes
no

yes
no

yes
no

yes
no

Lemmatisation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Entity rec
ognition (noun, number,
address,…)

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Terms clustering

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Synonym list creation

yes
no

yes
no

y
es
no

yes
no

yes
no

yes
no

yes
no

Number of representations

2

Variable

3

4


3

3

2

Lexical table
exportation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Lexical table Importation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Lexical table analysis








Classification methods diversity


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*
**
***


*

**

***

Automatic detection of classes

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Scoring

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Result presentation








Link to original text

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Graphical representation

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

yes
no

Reports Edition


yes
no

XML

yes
no

HTML

ye
s
no

Excel

yes
no


yes
no


yes
no

Excel

yes
no

HTML

Global appreciation of result presentation


*

**

***


*

**

***


*

**

***


*

**

***


*

**

***


*
**
***


*

**

***

Application types








Use recommended

Research

Customer relation

Art
icle

Research

Email, Web doc …

Marketing,
commercial,
customers

Social studies,
Surveys, political
discourses

Economic, scientific

Marketing, finance

Company
documentation

Observations






Industrial data,
investigations,
research papers,
human resourc
es



Analysis and
cartography of
documents

Themes structure
visualization



The
main characteristics of each too
l can
be
sum
med up as follows
:

1.

Temis Insight Discoverer
: Applies p
owe
rful linguistic techniques. It
allows for defining

a domain
with a
specifi
c lexicon

and themes/keywords classifications.

2.

SAS Text M
iner
: P
owerful data mining tech
niques (of paper, summary, forma
l
documents, investigations and forums).

3.

Tropes/Zoom
: F
ollow
-
up of text analysis. It represents results in different formats
(statistic
al and graphical).

4.

TKS
: Capable of analyzing
large volume
s

of text
. It is lar
gely used in research
laboratories
.

5.

NeuroNav
: Capable of analyzing
large volume
s

of text. It offers navigation and
classification environments (Web format).

6.

WordMappe
r: It provi
des g
raphical representation of clusters and their relations. It
supports several

data resources (Oracle, Lotus,
…). It can generate summary of
documents and report of analysis (with graphical and statistical information).

7.

SPAD/CRM
:
It has a
user
-
friendlin
ess interface. It su
pports several file formats
(MS
Word, Excel, data bases …)
.

It can gene
rate a rich report of analysis, including
graphi
cal
and statistical information
.


WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


23
/
56

5

Requirements towards the Knowledge Management system of the
VRL
KCiP network

Thi
s section gives a high
-
level specification of the services of the knowl
edge management system
that should

support the
VRL
KCiP community in the coming years.
Here we focus on the
functionality
of such a system.
It is taken for granted that the KMS will be
realized based on the
services of a web portal.
Hence, a technical specification is out of the scope of this section.

5.1

Position


We can distinguish the following main
functions

of
the
knowledge management tool

supporting
the
VRL:

1.

It should help

to

characte
rize and
identify people
and institutions
in the network
. I.e., it

should
facilitate

gathering
and
systematic ordering of information about the VRL, the
network members’
fields of expertise,
their relevant results

and their sources of
knowledge stored most
ly, but not exclusively, in documents.

2.

It should
provide a
cooperative work

environment for the VRL
network members.

3.

Finally, it should provide a framework
our industrial partners

can
use
later on
for finding
a network member
with
appropriate
expertise ov
er
a given
design or
production
related
issue
.

The first functionality must be operative
as soon as possible to facilitate the work of the VRL
duri
ng the project, while the last

is envisaged to be elaborated as an “end product” from which
the
industrial pa
rtners

will benefit.
Hence, in T
ask 26, and
specifically
in this deliverable
D51, we
focus on the first function.

We follow our original

hypothesis

that, in order to build a Knowledge Management System by
the use of a Document Management System, we are goi
ng to describe knowledge in documents.

We then propose with [Naeve et al., 2001]

to go from a
semantic web

aimed

at expressing metadata
that is machine
-
understandable to a

conceptual web

that is more comprehensible for humans. “It
gives the user a clear ov
erview of the subject area (=context) while at the same time allowing the
exploration of its various forms of content. Incorporating web resources [or documents] as
content is done by associating concepts with occurrences in resources”

[Naeve et al., 2005,

p.
358].

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


24
/
56

Further, if the
Knowledge Management System

is to survive, it should be dyn
amic and has to
facilitate adding new and removing old information.

Some of the misconceptions regarding
metadata (i.e.
,

the topic
-
categorization) is
just
that it is obje
ctive, static and has logically defined
semantics. Even the production domain is changing


with new technologies and interdisciplinary
methods


and there is a need for supporting dynamically evolving metadata over multiple
vocabularies and taxonomies.

A
nother question is whether it should be possible to relate the VRL work to results from other
communities, published within the internet. If we want this to be a persistent bas
is for wider
collaboration,
our conceptual map
must be harmonized with other web
-
based taxonomies.

5.2

System architecture

We propose to structure the system in three layers:

1.

Plain documents
: articles, handbooks, technical reports,
patents, demos
, etc.
T
his layer
evolves constantly
.

2.

Essays/items
:

which contain knowledge from the author
’s

point of view, and links to
different kind of documents (articles, illustrations, demos …). The essays
wh
ich are
themselves documents
typic
ally

index documents. This layer evolves constantly
.

3.

Conceptual map
:
related to documents,
especially to essays,
rel
atively stable. By
browsing this conceptual map, the user will find some interesting items, and will then
access to essays, and finally to plain documents. This layer is in
strong
relation with work
done in IA
-
WP1 and other tasks of JRA
-
WP3: it is composed

by using the taxonomy
developed by IA
-
WP1 as well as the conceptual maps of some technical fields (design and
virtual prototyping,
processes, simulation).

There are then two
main
strategies to find the right person to work with or to find the work done
by

people in the network:



by
browsing

the conceptual map,

and



by
searching

for documents or topics described in essays.

5.3

T
ypical u
ser processes of the Knowledge Management

S
ystem

Here
we suggest

typical
user processes for the network

s Knowledge Management
System and
present them as so
-
called
use cases
.



WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


25
/
56


Figure
2
.
VRL KCiP Knowledge Management System

Use Cases


We propose to put together the function
s

which allow

defining a profile and to take out a
subs
cription, and the function which allows describing essay(s), in order to make users aware of
their content responsibilities.

5.3.1

Use Case


Creating and updating an individual profile

On the home page of the VRL KM tool, the user can choose to create his/her p
rofile, or to
update it.

When the user creates his/her profile, he/she completes a form containing
information on institution, research domain, research topics.

The user is then able to create one or several essays on his/her work in relation with
VRL KCi
P

network. Each essay describes a short description of his/her knowledge background and research
interests. To create an essay, the user will complete a form with:



t
itle
,



a
uthor
,



l
ead
,



b
ody
,

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


26
/
56



links: t
he user is able to upload documents
,



related concepts,



keywords
, and



illustrations.

The user can then take out a subscription to some topics of interest. He/she will then receive
news on this topic in the VRL KM tool by email.

5.3.2

Use Case
-

Creating and updating an institutional profile

Each institution of the VR
L will choose one of its members which will be in charge of creating
and updating institutional profile.

On the home page of the VRL KM tool, the institution
representative can choose to create the institution profile, or to update it.

When the user creat
es the profile, he/she completes a form containing information on
institution, research domain, and research topics.

The user is then able to create one or several
essays on general work done in his/her institution in relation with
VRL KCiP

network. To cre
ate
an essay, the user will complete a form with:



t
itle
,



institution
,



lead
,



body
,



links (t
he user is able to upload documents
)
,



related concepts,



keywords
, and



illustrations.

5.3.3

Use Case


Querying
the

system

On the home page of the VRL KM tool, a user can
choose to query the system. A query form
appears where the user defines his/her search criteria.

Further on, the user may want to be able
to search for topics from different points of view, e.g. based on different keywords.

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


27
/
56

The results are classified by ty
pes: essays or documents, and the user can choose to classify them
by relevance, institution or publication date.

5.3.4

Use Case


Browsing

the conceptual map

In this case, the user works on the

first layer of the system.

On the home page of the VRL KM tool, use
rs see the conceptual map, and are able to browse
this map. The conceptual map is the fi
rst layer of the system, and is composed of
the taxonomy
proposed by
IA
-
WP1.

The user is viewing a particular index, and is navigat
ing around it and its sub
-
indices

in

search of
int
eresting items. For instance, here indices belong to
:



c
ountry
,



i
nstitution
,



research domain
,



product life cycle phase
.

Indices of portion of the index that the user finds interesting are displayed: title, date and authors.
This information c
omes from the essays (see use case Creating an individual profile). The user
can select one of the essay
s
/item
s

to visualize it.

5.3.5

Use Case


Visualizing

an item/essay

In this case, the user works on
the second layer of the system.

After browsing the concept
ual map, the user has found a list of items/essays. He/she has
selected one of them, and the system displays it. The user can see all the forms which were filled
in: title, author, lead, body, keywords, and links.

He/she can choose to click on a link whic
h will open a web page
, a document, an illustration.
Hence, t
his
will lead to
the third layer of the system.

5.4

A
data

model for the Knowledge Management System

Here

we pro
pose an initial data
model for
the KMS.
The UML
-
lik
e schemes detailed below
specify

a p
ossible str
ucture
of the database

for

the
most essential components
of the
three layers
of the
KM
system. This model should be

regarded as a starting point for further development.



WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


28
/
56

5.4.1

Keeping track of researchers and institutions

The “atomic” element of our

network


as of a
r
esearch institution


is the
individual researcher
, as
shown
in Fig. 3
. A researcher can be affiliated with a group of other researchers (or several
groups). Since research groups and institutions incorporating several groups at various

levels
exhibit a similar behavior, this simple model unifies all such groups at all possible
hierarchical
level
s

under the term “institution”. One given instance can then be a part of other instances of
“institution”


a research group formed in cooperati
on with several institutions can p
oint its “is
-
part
-
of” relations

to several other instances of the class.



Figure
3
. Researchers and institutions


5.4.2

Mapping of concepts

Another impo
rtant component of the concept

map is a systemati
c directory
of the different
concepts

related to various domains of science, application
areas, e
conomic and social
environments

etc
. While it is clear that taking specific problems or problem groups for atomic
elements would result in a more accurate mode
l, the associated amount of work would not justify
its use in the VRL
-
KCiP project. Therefore, as shown in
the sequence of Figures 4, 5, and 6
,
having a somewhat redundant scheme with both hierarchical concepts and
keywords can be
taken as a fair
, robust

c
ompromise, keeping in mind that keywords can hardly give complete and
unambiguous coverage of the domains in question but still provide at least a few simple
-
to
-
use
entry points. Therefore, an agreeable structure for this part of the database coul
d look as

shown
in Fig. 4
:

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


29
/
56



Figure
4
. Hierarchical concep
t
s


Among
concepts

(whose subordinates
may be
subconcepts

of various levels, just as institutes
mentioned above), two relations are
the
most important:



Is
-
a

relation: This is the c
l
assical
sub
-
super concept

relation. An example is shown below:


Figure
5
. Relation within a concept

and its subconcepts




Contributes
-
to

relation:
Typically, new concepts are

formed around problem grou
ps
which require expertise abo
ut
traditionally rather disjoint

concepts
. An example
with
robotics is shown below:


WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


30
/
56


Figure
6
. Example for the
contributes
-
to

relation

5.4.3

Mapping documents, including essays

The third main components in the VRL compet
ence map are th
e
documents
, like
publications,
technical reports and various other information resource
s

grouped under the “
Concept
” class.
Note th
at fixed
-
form essays belong also to this class.
One such document
can be part of another
one (such as a paper in a

proceedin
gs volume), and can refer to
other works among its references,
as shown below:



Figure
7
. The Documents class


Note that Essays
are Documents with special
“Indexes”
relations that help to link various plain
documents (like articl
es, technical reports, etc.) to them.

5.4.4

Connecting the three expertise groups

Having outlined the three major groups within the
conceptual

map (researchers, concepts and
documents), what remains to be specified are th
e relations between these main
classes.
D
ocuments are connected to researchers via the “Has
-
published” relation which also specifies
how the given researcher has contributed to the creation of the given product (e.g., author, editor
etc.)

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


31
/
56

A document may have keywords (shown through “Has
-
keyword”)

which bel
ong to a given
concept
. Also, it may prove necessary to make direct connections between a document and a
concept which is realized by the “Deals
-
with” relation.


Figure
8
. Relations of
documents
, concepts and keywords


I
t may also be possible that no
document

of a researcher
related to a specific concept

is present
in the database, yet the researcher is
interested in the concept

to some degree. This may be either
due to the researcher having not completed yet any product
or publication in the field of
question, or due to the database being incomplete (which it will indeed always be, as new
information will always be added to it as time passes). To handle this case, a direct link may be
established between researchers and
c
oncepts
. The resulting scheme is as follows:



Figure
9
.
Direct r
elation between researcher and concept


WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


32
/
56

From the above elements, we build up now the complete data model:



Figure
10
. The complete data mo
del


The above model should be regarded as a proposal for a minimalist data model of the Knowledge
Management System of the network. Certainly, it should be developed further to include
authorization and access control, versioning history, to name but a fe
w issues. This development
should go hand in hand with works in IA
-
WP1 and WP2. However, having a clear
-
cut model is
essential
for constructing,
running,
using and maintaining the system.

5.5

Supporting cooperative work

We
can but
note
here
that use processes

for supporting cooperative work are important,

but are
out of the scope of this deliverable. The specification of these processes belong
s

to
JRA
-
WP3

T
ask 27

and will be described in
deliverable
D
52.

WP

JRA
-
WP3




Contributing tasks

T26

Version

1.0

Responsible organisation

MTA SZTAKI

Last update

30 May, 2005



Deliverable No 51.


33
/
56

6

Conclusions

In this document and the related deliverab
le, D50 we argued that setting up the knowledge
management system of the network is a knowledge management process itself. Based on the
lessons of an extensive survey
-
based and a subsequent personally directed knowledge acquisition
process, in this documen