KNOWLEDGE MANAGEMENT IN INFORMATION SYSTEM

DESIGN AND DELIVERY PROCESS

An Application to the Design of a Legal Information System

Ovidiu Vasutiu

a,b

, Youssef Amghar

, David Jouve

and Jean-Marie Pinon

Laboratoire d'InfoRmatique en Images et Systèmes d'information – INSA de Lyon, France

Caisse Nationale des Allocations Familiales – Cnedi Rhône-Alpes, France

Keywords: System design process, legal document, document retrieval, document consistency, UML, XML.

Abstract: Nowadays information systems are more and more important for all types of organizations. To deal with the

complex technologies available, IT specialists use proven best practices inspired from more comprehensive

process frameworks for software and systems delivery or implementation and for effective project

management. Methods developed to support theses processes produce many different heterogeneous

resources (design documents and models, planning, project prototypes, etc...). Due to the continuously

changing reality, designers will always have to consider new user and stakeholders requirements and go

back to the starting design case for an update. In this paper we present an organizational and technical

infrastructure for a collaborative design process management system which automates mechanisms to assure

the coherence and consistency of these heterogeneous and continuously updated resources..

1 INTRODUCTION

Organizations are turning out to be more and more

dependent on Information Systems. At the moment

where IT technologies are becoming more complex,

where many frameworks and architectures models

are available, the need of tools to support the

Information System design, delivery and

implementation process is very important. This

design process take in consideration large teams and

deal with many different skills and profiles: users

and stakeholders define the functional requirements,

software and technical architects elaborate the

system architecture, developers build it, designer

package it, etc… Each one of these actors provides a

different view and knowledge on the system. Their

functional, technical and organisational knowledge,

while representing different aspects of the same

system, will be expressed by heterogeneous

resources: documents and models in different

formats, and, finally, by applications’ source code.

Methods such as RUP (IBM, 2007), MSF

(Microsoft, 2007) or Extreme Programming

(Larman, 2003), have been developed to support this

process using different phases (i.e. inception,

elaboration, construction, transition) each one

producing different deliverables: documents,

models. Due to the complexity of the systems,

incremental and iterative approaches are used. The

system is then delivered into sets of iterations

producing or updating deliverables (design

resources, documents, models). Information Systems

need to follow, to adapt to a continuously changing

reality. Therefore the system’s knowledge is

building all the way during the design and

implementation phases making it very important to

assure the coherence of the design resources.

Furthermore, in many domains (legal organisations,

banking, e-government, etc.) it is also very

important to guarantee the traceability all the way

through the process; each component, each update

has to be related to a user or functional requirement.

In this paper we propose a system based on

knowledge management to control and to guaranty

the coherence and the consistency of theses

heterogeneous resources (documents and models)

and the information system.

The work reported in this paper has been carried

out in the context of the Caisse Nationale des

Allocations Familiales (Cnaf), the French Family

Benefits Office, whose main mission is applying the

law in the process of evaluating each family’s

allowances. Therefore, our proposition was

implemented in this particular domain and

organization. However, it is not restricted to the

492

Vasutiu O., Amghar Y., Jouve D. and Pinon J. (2008).

KNOWLEDGE MANAGEMENT IN INFORMATION SYSTEM DESIGN AND DELIVERY PROCESS - An Application to the Design of a Legal Information

System.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - DISI, pages 492-496

DOI: 10.5220/0001708204920496

 SciTePress

legal domain. It contains generic elements that can

be either directly reused or extended so as to

represent specifics of various domains of interest.

2 MOTIVATIONS

The legislation of our organization’s domain is

particularly complex. The organization has to deal

with a massive amount of legal documents (more

than 100,000 references) and numerous legal

updates (over two legal updates every week).

Considering the important number of beneficiaries

and the large amount of information to be processed,

the workload of this organization overwhelms

human capacities forcing it to use the computer’s

assistance, a Legal Information System.

For any legal sub-domain, the set of relevant

legal documents is partially structured (Jouve et al,

2001), documents can be laid out in a pyramidal

structure according to their legal importance, Figure

1. The highly operational degree documents are at

the lowest level of the pyramid. They are used for

the Legal Information System design.

eg al wei ght

ocument kind

Constitution

Statute

De cree

Mi ni st eri al circ ul ar

Internal reference text

Technical instruction

Or igi n

Parliament

ational Assembly

Council of minister

Gove rmen t de

Or ganiza ti on le vel

oftware implementation of

the legislation

Conception models

Func tional documents

Development

User documentation

IS Com p 1

…

IS Co mp 3 IS Comp 2

…

IS Comp.

…

IS C omp

…

Figure 1: Legal and design resources hierarchy.

Due to the legal context of each country,

legislation is constantly added, deleted or modified,

like an informational flow in perpetual evolution.

One change at a given level of the legal pyramid will

have repercussions to all beneath levels until the

lowest level is reached, affecting the information

systems components that are directly and strongly

connected to the legal domain.

Information Technology specialists use

comprehensive process frameworks, providing then

with proven best practices for software and systems

delivery and effective project management. For

example the RUP - Rational Unified Process

proposes a project design life cycle in 4 phases:

inception, elaboration, construction, transition. In

each phase different actors (users, designer,

architects and trainers) take part producing

heterogeneous resources (analysis documents, UML

(OMG, 2006) models). This resources are

interconnected and they express the same knowledge

but on different points of view.

Figure 2: Phases and heterogeneous deliverables.

When user requirements change, the update has

repercussions to all others design documents. There

are solutions for document management systems and

collaborative work; others solutions for UML

modelling; and even for requirements management,

but not many solutions control the coherence and

consistency of both kinds of resources. In this paper,

we propose a logical and technical architecture for

the design resources management and production

environment of our organization.

3 STATE OF THE ART

3.1 Structuring Documents

XML(W3C, 2006) describes the logical structure of

the documents using mark-up, enabling both an

explicit representation of the given information and

a description of relations between modelled

information and documents sub-structures.

Furthermore the benefit of mark-up languages is that

they allow a clear separation between data and its

representation.

XML documents are stored in a native XML

databases which uses the XML document as the

fundamental unit of storage having better

performance on most of the rich content document

queries (Runapongsa et al., 2006).

KNOWLEDGE MANAGEMENT IN INFORMATION SYSTEM DESIGN AND DELIVERY PROCESS - An Application

to the Design of a Legal Information System

493

3.2 Document Integrity

Research projects in literature address document

integrity in a document repository based on link

integrity check focusing their work on the intranet

environment (Amghar & Chbeir, 2003), others

(Ashman, 2000) summed up all measures in two

categories. (1) preventive solutions : forbidding

changes, versioning, embedded links, external links,

link attributes and (2) correctives solutions : forward

mechanism, Universal resource names, relative

references, utilization of the paradigm of agents.

3.3 Knowledge Representation

Many scientific works managed to accomplish

models for the representation and manipulation of

legal documents (Wright, 1963). David Jouve (Jouve

et al., 2003) addressed the issues caused by the

constantly changing legislation proposing a solution

to the need to efficiently control a legal domain:

impact studies, incoherence and conflicts detection.

Others related works interested on the

management of human knowledge within the design

process, the collaborative work and knowledge

sharing (Rodríguez et al., 2004) and the

organizational memory (Rus and Lindvall, 2002).

We propose to use knowledge management for

consistency and coherence control mechanisms

within heterogeneous resources.

4 DESIGN PROCESS

MANAGEMENT

ENVIRONMENT

Our answer to the issues presented here is an XML

based solution for a production and management

environment for information system design

resources. The environment developed at the Cnaf

has an N-layer type of architecture.

4.1 Meta-Model of the DMIS

The design document referential embeds a set of

Document collections. Each collection is provided

with a set of descriptors defining all the types of

contained resource: document types (i.e. legal

document, analysis, design, etc.), asset types,

reference types… The collection meta-description

contains taxonomies, docflow definitions and

references types.

4.2 Basic Features

a) Document format and structure

To avoid redundancy, two structures are used:

the document persistence structure - content stored

at only one place, enabling document archiving,

indexing, and retrieval -, and the document

presentation structure (user ended documents

containing richer and more detailed information, an

aggregation from different persistent documents).

All documents respect a generic structure having

two top elements: the meta element (all document

meta-data: document type, ID, version, revision and

edition identifiers, origin, state and docflow history)

and the content element (contains the information).

The system uses a native XML database to store

persistence structures. For backward or external

interoperability raisons it can also wrap in an XML

structure any other content.

b) Version and revision management

The stable reference documentation is stored

with the “reference” version in the reference space.

When a new user requirement occurs a new editing

workspace is created containing documents to be

added, modified or deleted (Figure 3) in order to

respond to that requirement. The new workspace is

created starting from the reference space therefore in

each workspace we find new versions of the

reference documents. This new versions follow

theirs document editing life-cycle leading to

different revisions. At any time, editors can

synchronize, using a merge tool, their working

version of the document with the reference version

of the document (the public version of the

document).

Figure 3: Multiple parallel document editing workspaces.

Once a working version of the application is

validated all the modifications will be integrated in

the reference space, updating it. This situation

sometimes leads to multiple parallel versions of the

ICEIS 2008 - International Conference on Enterprise Information Systems

494

application having parallel versions of the

application’s documentation.

c) Resource identification

Knowledge is scattered all over the document

corpus thus preserving link integrity of the document

repository is a crucial problem. A solution to

precisely addressing a resource is the use of

Universal Resource Names or URNs. A revision of a

document is precisely identified by the unique

document identifier, its version and revision number.

It is referenced by a storage independent URN

formed as follows:

collection-name : resource-kind :

resource-type : resource-id

[: version : revision]

Where the resource-kind describe the kind (

document, image, an external resource), resource-

type (data description document, class description

documents, legal document, etc…); resource-id: is

the unique identifier given by the database.

This identification allows us to have a unique

reference model where a resource can be moved

without affecting the references pointing it. URN

resolution rules will identify the physical document,

taking in consideration if the version and/or revision

are specified or not. Documents with the current or

reference version and those in the last revision are

privileged. The advantages of this link resolution

algorithm are that a link can be defined to always

point out to the last revision of a document even if

the target document evolves, or it can be strictly

forced to point out to a specific revision number,

even if obsolete.

d) Document Addressing Model

A reference is encoded as an XML element with

a set of attributes that is mapped to the URN of the

target:

<ref col="sair-daf" kind="docobject"

type="tache" res="00000003"

[ver=”REF” rev=”0001” nodes=”id1 id2”

ref-type="originContent"]/>

The col, kind, type, res define the URN of the

referenced resource. The optional nodes attribute can

address a specific sub-fragment of the document.

Referenced document’s label, description, type are

retrieved. Sub-fragments identified by the nodes

attribute, content defined by the ref-type are

included in the presentation of the referencing

document. All the information concerning the target

document is stored within so that any update is

automatically available. We can identify two

categories of reference types: (1) simple reference: a

document points out to another document implying a

simple link, (2) specialized reference: attribute ref-

type is present defining the semantic of the link. This

semantic defines the transformations to be made

during loading to build the presentation document.

e) Docflow

Each document at a given moment is associated

to a state in its lifecycle (specific to each document

type). User contextual menu functions are defined

according to user’s authorizations and document

state. An internal docflow engine listens to external

(function activation, service call, etc...) or internal

(document transitions) events that might (if all

conditions are validated) launch a transition to

another state. Actions can be executed on in/out

transitions.

f) Information retrieval

The retrieval mechanism is very flexible

allowing users to specify a search expression

(containing multiple keywords or phrases) and/or

document expected metadata (identifier, document

type, current state, date).

4.3 Design Document Drafting

Functionalities

a) Unique referential

The resource addressing feature and link

resolution algorithm allow us to reduce redundancy

without losing any valuable performance. A link can

either point to the last revision of a document, or it

can be strictly forced to point out to a specific

revision number, even if obsolete. So documents in

parallel editions workspaces (different versions) can

be referenced without any interference. Using the

“ref-type” attribute, we can define meaningful links

between two documents (i.e. origin of the document,

knowledge dependency or similarity) or how content

from a document is included in the presentation

structure of another document.

b) Drafting process management

Responsibility in the design process is often

shared between many actors. Docflow mechanisms

allow us to define the automation of the drafting

process, the roles of each participant, the flow of

documents and information, in whole or part, from

one decisional instance or state to another for action.

c) Knowledge and information traceability

Meaningful links allow us to trace the usage of

the knowledge and its propagation while meta-data

allow us to trace evolutions, contributors,

contributions and updates of the documents.

Software engineers use a specific process to create

the application components needed to respond to

each user or functional requirement. The history of

KNOWLEDGE MANAGEMENT IN INFORMATION SYSTEM DESIGN AND DELIVERY PROCESS - An Application

to the Design of a Legal Information System

495

each document can give us the information needed

to retrace back all the interventions on the document

to the linked document fragment expressing that

requirement. Therefore we can define a new

specialization of the reference relation, the “user

requirement implementation relation”.

d) Dependency and impact analysis

Due to traceability in our system we can identify

the documents that are in a way related to a

requirement text and that need to be updated with

each evolution, giving an estimate of the impact

(Vasutiu et al., 2006). In our case as Legal

Information System design documents are related to

the law, it allows us to link legal requirements back

to corresponding design artefacts and code.

5 DISCUSSION AND

CONCLUSIONS

The document and model management system

presented here and implemented at the Cnaf, while

adapted to the particularities of the legal domain,

responds to the generic demands expressed.

For instance coherence is assured by the

reference and traceability mechanisms. References

can allow us to avoid redundancy, often present

especially in legal documents. Knowledge

equivalence reference allow us to identify

documents, although different but expressing the

same knowledge. The complexity of the knowledge

transformation within the design process is managed

by the docflow and version/revision control features.

To specifically develop this infrastructure took

over one year. The system was tested on a 30,000

design document collection which represents almost

30% of the document collection. All the feed-back

received, after a user validation phase, shows that

even if the new mechanisms implied more

constraints in the work process, they help to reduce

the risk of error. Traceability helped users to

precisely identify the documents that need to be

updated and version/revision control allowed then to

constantly be able to control changes and evolutions.

Therefore the next perspective of our research is

to estimate the whole area of impact of a user,

functional or legal requirement using a dependency

and impact analysis on both documentary area and

on the information system.

REFERENCES

Amghar Y., Chbeir R., (2003). Advanced Management of

Documents Integrity in an Intranet Environment,

ACS/IEEE Conference: International Conference on

Computer Systems and Applications, Tunisia, 2003

Ashman, H., (2000). Electronic Document Addressing:

Dealing with Change. ACM Computing Surveys, Vol.

32, No.3, September 2000

IBM International Business Machines (2007) Rational

Unified Process, Retrieved September 15, from

http://www-306.ibm.com/software/awdtools/rup/

Jouve D., Chabbat B., Amghar Y., Pinon J-M. (2001).

L’archivage des documents réglementaires : maîtrise

d’une matière première spécifique. Document

Numérique, Vol. 4, (pp. 315-329).

Jouve D., Amghar Y., Chabbat B., Pinon J-M. (2003).

Conceptual Framework for Document Semantic

Modelling: an Application to Document and

Knowledge Management in the Legal Domain. Data

& Knowledge Engineering, Vol. 46, No. 3, (pp. 345-

375).

Larman, C., V. Basili (2003), Iterative and Incremental

Development: A Brief History, Computer (IEEE

Computer Society) 36 (6) pp. 47-56.

Microsoft (2007) Microsoft Software Fondation. Retrieved

September 15th from: http://www.microsoft.com

/technet/solutionaccelerators/msf/default.mspx

OMG, Object Management Group, Unified Modeling

Language Specification version 1.5, formal/2003-03-

01 Retrieved September 15th from:

http://www.omg.org/

O.M. Rodríguez, A.I. Martínez, J. Favela, A. Vizcaíno,

and M. Piattini, 2004, Understanding and Supporting

Knowledge Flows in a Community of Software

Developers, LNCS 3198: 52-66.

Runapongsa K., Pate, J.M., Jagadish, H.V., Chen, Y., Al-

Khalifa, S., (2006). The Michigan benchmark: towards

XML query performance diagnostics, Information

Systems No. 31, (pp. 73-97)

Rus, I., Lindvall, M., 2002, Knowledge Management in

Software Engineering, IEEE Software, 19(): 26-38.

Vasutiu O., Amghar Y., Jouve D. (2006) Gestion des

changements et étude d’impact dans un système

d’information réglementaire». XXIVème Congrès

INFormatique des Organisations et Systèmes

d’Information et de Décision, pp. 1007-1022, (2006).

von Wright, G.H. (1963). Norm and Action: A Logical

Enquiry, International Library of Philosophy and

Scientific Method, Routledge and Kegan Paul,

London.

W3C World Wide Web Consortium, (2006) Extensible

Markup Language (XML) 1.0 (Fourth Edition) W3C

Recommendation 16 August 2006. Retrieved April 10,

2007, from http://www.w3.org/TR/REC-xml/

ICEIS 2008 - International Conference on Enterprise Information Systems

496