Using Electronic Structured Document as an Interface

Using Electronic Structured Document as an Interface for Knowledge Based System Samuel PARFOURU1,2, Sylvain Mahé2, Manuel Zacklad1 1

Institut Charles Delaunay, Université de technologie de Troyes (France), FRE CNRS 2848 Laboratoire Technologie de la Coopération pour l'Innovation et le Changement Organisationnel (Tech-CICO), ([email protected]) 2 Département STEP - Simulation et Traitement de l'information pour l'Exploitation des systèmes de Production Groupe SDTI - Systèmes Dynamiques et Traitement de l'Information EDF Recherche et Développement, 6 quai Watier 78100 Chatou, France (firstname.surname @edf.fr) ABSTRACT Development of Knowledge-Based System (KBS) represents a crucial point in enterprises because expert knowledge and know-how are an important immaterial capital. Preservation, sharing and enhance of this capital is essential for enterprise activity. In this way, integration of KBS in enterprise information system represents an essential challenge. In this paper we suggest that is really important to take into consideration particularities of KBS. As a consequence deployment of KBS is not limited to connect it to enterprises information system. Our approach is to consider that Knowledge Base Systems correspond to systems that mediate collective activity of networked actors in an asynchronous and delocalized cooperation. So, we think that classical Human Computer Interface could not respond to that. In fact, the problem is to fit knowledge representation to a context that relies on users and activities that we want to support by KBS and also support cooperation. In this way, we propose an electronic structured document approach of Human Computer Interface. Using document as an interface conveys an important flexibility in structuring and formatting of data that is essential in the restitution of knowledge representation. Moreover, document with his media dimension constitutes a good artefact in collaborative activities. Finally, we think that using electronic structured document as human computer interface represents the opportunity to transform the user as an actor in the restitution process of knowledge representation. The document as an interface is relies on the Action Document notion [9]. Keywords: Electronic Structured Document, Knowledge Management, Human Computer Interface, Restitution process, Data capture process, XML 1.

INTRODUCTION

In the past fifteen years a significant amount of work has been devoted to the development of Knowledge-Based System (KBS)[24][27]. This kind of system represents a key point for the activity of enterprises and organizations because they contain a part of expert knowledge and know-how [25][26]. On the one hand, the preservation of expert knowledge and know-how is a major issue because knowledge corresponds to an important immaterial capital of an enterprise. On the other hand, it is a crucial point to share information from this system and to enhance the value of it in enterprise activity. In this way, integration of KBS in enterprise information system represents an essential challenge. However, it is important to take in consideration that problem is not only to connect and to integrate KBS to enterprise network and produce data flow supported by web services technologies for example. It is really important to take in consideration difficulties that convey KBS and particularly the restitution process of a knowledge representation.

In this paper, we are interested in the definition and the restitution of knowledge representation stored by KBS. Knowledge representation doesn’t correspond to classical computer data, which is easily possible to define precisely data type, data format and so on. The problem is not only to provide literally the knowledge representation in a standard form, but it is to fit it on the context and the activity that it should support. In this way we propose an approach of the restitution of knowledge representation based on electronic structured document. We argue that document could become the interface between systems, especially KBS, and users. In fact, the document could introduce a real flexibility in the restitution process, and also in the data capture process. Moreover, using document as an interface conveys a part of autonomy with the system where knowledge representation comes from. That is very interesting in the deployement of services where we need to connect differents systems and exchange data between them. By using XML technologies that is entirely compliant with web services deployment. Our study is supported with a real use case. This use case relies on the design and deployment of a

Knowledge base System called “Diagnosis guide”. The Diagnosis guide is designed to capitalize and to share the expert knowledge and know-how about the maintenance of hydraulic building in the field of power supply. This paper is organized as follow. Section 2 introduces Knowledge Base Systems and particularly the design of this kind of systems. Section 3 presents our approach where we propose to use Electronic Document as Human Computer Interface. Section 4 introduces more practical aspects in the use case of the Diagnosis Guide. Finally we conclude. 2. KNOWLEDGE BASE SYSTEMS (KBS)

Knowledge Base Systems are designed in order to meet two complementary objectives [3][7]. On the one hand, the capitalisation of know how of some specialists of a particular domain that we are interested in. On the other hand, the restitution (writing data) of a knowledge capitalisation with “human computer dialog” that endeavours to help users in solving a particular “problem situation”. Generally, we can observe three main difficulties in the design of Knowledge Base Systems (KBS): • The first difficulty is the problem of finding the clear way to formalize the expert know how. The formalization must be a compromise between general and specific level to become an efficient artefact for final users; • Secondly, we are confronted by user appropriation problem. This problem is in relation with the restitution of expert knowledge to the user providing contextual help that is useful and usable; • Third difficulty is continually updating the knowledge base as it evolves. For a long time, Artificial Intelligence has promoted the design of programs that are able to simulate expert reasoning and this main vision has widely led to expert systems development. This idea was related to the search for the correct expression standard and often lead to produce very prescriptive systems [24]. Another vision, promoted especially by “COOP” conferences, was a vision of cooperation between system and user permitted by the program. In this way, the key point is to propose a restitution easy to understand and apply related to appropriation concept. The approach that we suggest in this paper, mostly inspired by Computer Supported Cooperative Work (CSCW) [4][15] and Information Science, is to consider that Knowledge Base Systems correspond to a system that mediate collective activity of networked actors in an asynchronous and delocalized

cooperation. This collective activity is effective during all the life cycle of the KBS. Thus the actor network is composed of knowledge engineer, experts and final users. Knowledge engineer leads the knowledge management project and particularly the design of knowledge representation. Experts try to explain their knowledge and the know how about an object or an activity. Generally, we contact several experts. In Diagnosis Guide example, we contact experts from three skills : civil engineering, hydromechanics and control command. Each field introduce specific knowledge, specific vocabulary and specific point of view. Final users of KBS try to define the need particularly about the restitution of knowledge. All actors must collaborative to define a KBS and a knowledge representation enhance able. 2.1 The knowledge base system design

We have looked at some KBS designs. We noticed that this is influenced by a tradition of information system particularly about Human Computer Interface design. It is often a predictive design (classical method), the result is that it is frequently a predictive dialog between human and computer especially in data capture interface. We can observe many formalizes that imposes a strict procedural data capture processes [16]. This data capture process is often according to computer structuring or algorithm implantation: the design is focalized on technology [14]. The Human Computer Interface is most designed as a way to access to application (Knowledge Base System) functions and not as an artefact that could support real user activity. We argue that is difficult to meet the entire specificities of Knowledge Management (KM) activities by this design method. Thus, the knowledge capitalization is not a determinist process and is mostly a creative activity: out of capitalization itself, it seems difficult to anticipate information that will be effectively identified, their structuring and also the data capture process (it is often dynamic and fuzzy at the beginning of capitalization process). Moreover, people that take part in the capitalization introduce fuzziness: the think of an ideal user must be given up interface design [14]. It is the same situation when users use KBS, because it will modify user activities [17][18][19]. So it’s still really difficult to anticipate with accuracy the information need, information layout, information shape or also the way this information will be enhanced. In this context, if the KBS sets too strict logic, leading only to one point of view, it couldn’t be compliant with the need to offer problem-solving

space [6] in step with our position. In this way, firstly, it seems to be necessary to provide flexible data capture that corresponds to efficient use of media space for creative activity. Secondly, it seems to be essential to try to adapt the restitution according to user variability. This variability depends on expertise, skills, and origin field for example. Therefore, we suggest that is necessary to go away from classical Human Computer Interface Design. Generally, it is centered on the system and the technologies [14][20]. So we propose to use electronic structured document to take place of the interface between users and KBS. 3. USING ELECTRONIC STRUCTURED DOCUMENT AS HUMAN COMPUTER INTERFACE

We suggest an electronic document-based approach of the KBS. In fact, we argue that the KBS could be seen as an interactive electronic document that mediates knowledge representations between actors [21]. Thus, the knowledge engineer, experts and final users could take successively the role of author, editor or reader in front of the system and the restitution. The objective is to place the user as an author of his own interface with the system. Most of the difficulties that we described previously are related to the issue of the document as a “form” and as a “sign” as explained in [8] and so a visible part of KBS: Human Computer Interface (HCI). The idea that we consider the notion of document is in accordance with the vision of the HCI [10] [4] [2] that can be thought as a form of mediated communication between participants of knowledge management project. Thus, we can take advantage of using structured document as an interface in this context. First, during KBS development, it is possible to go from a predictive design activity (model design, knowledge representation language…) to a writing activity: the data capture process is on a declarative manner. Indeed, logical structure (tag definition and tag hierarchy) and content (content of tag) could grow together progressively during capitalization process. We are close to Extreme Programming approach [28]. The document becomes a design space where it is possible to fit the compromise between general and specific expression of knowledge or to combine them. So we have the opportunity by the handling and the transformation of documents to propose to users general or specific vision. In this way, it is possible to process structured documents to associate information from different

documents, to organize or to filter information in different way and to represent information in different modalities. The structured document represents a flexible media because it proposes a separation between content (“sign”) and representation (“form”). The idea is to take advantage of this separation to design HCI like Model –View – Controler approach [22]. But in comparaison with a formalize (classical Human Computer Interface), it is easier and cheaper with document (on the development cost aspect) to fit the restitution of knowledge representation to a specific context like data capture of knowledge or integration of knowledge representation in enterprise activity. Thus in the example of Diagnosis Guide, the Knowledge Base is composed of structured documents that associate information from different skills. In the capitalization process, the knowledge engineer produces specific version of the Diagnostic Guide for each skill and send it to experts for a review (annotative process for example). This specific version constitutes a document that mediates the cooperation between knowledge engineer and expert in the design of knowledge representations. That corresponds to a personalization of the content (the sign) that contributes to a better cooperation between actors. Each of documents constitutes an Action Document [9]. Personalize representation (form) is also an essential point. First, to fit representation in accordance with the user and that is able to be understand [4][5]. Then in the Diagnosis Guide example, we can observe that the knowledge engineer takes advantage of form personalization in the review. For example, he emphasizes data on what he want to concentrate on. We have shortly presented the electronic structured document as a really flexible interface. Then, we propose now to describe the using of electronic structured document in the real use case of the Diagnostic Guide. 4. ELECTRONIC DOCUMENT SYSTEM FOR KNOWLEDGE MANAGEMENT IN A REAL CONTEXT : USE CASE OF DIAGNOSTIC GUIDE

We propose now to describe an Electronic Document Device that we have designed to support the development of the Diagnosis Guide. This device is an association of tools that allow writing and handling of structured documents in XML format. We use XML editors (source editor, wysiwyg editor), office application (Word, Excel…) and Internet Browsers for data capture and restitution processes. All applications are connected to the web development framework Cocoon (http://cocoon.apache.org/) that we use to handle and

publish electronic documents. This server is very flexible because it is based on XML and Java technologies and it is easy to connect it to existing KBS or any system compliant with this kind of technologies. For example, this web server is very convenient for the deployment. In the publishing process, electronic structured documents become an interface between user and KBS. The web server cocoon is able to produce lots of formats that are very convenient to adapt restitution of knowledge representation to real context. For example, we can propose XML, HTML (Figure 3), RTF, PDF (Figure 1) or also SVG format (Figure 4). For each format, it is possible to produce multimodal restitution that combines text and graphic… Those possibilities allow to fit restitution and also data capture to real using contexts.

Figure 1. The Diagnosis Guide in PDF Format For the Diagnosis Guide, one of problem is to integrate it into an existing IT-infrastructure and takes into consideration available software. Users are used to some specific softwares : it is important as well to take in consideration this parameter in the deployment of Diagnosis Guide. So, in this use case, we broadcast knowledge representation by Internet browsers. We propose two main formats. First, HTML that is particularly fitted to browsing the Diagnosis Guide. Second, PDF format is also a format that proposes good browsing functions. Moreover, PDF is a format that produces a concrete form of a document and easy to print (Figure 1). It is essential in a real context where it is not always possible to have a computer on the work ground. We have also produced RTF format that is compliant with Microsoft Word that represents a good opportunity to adapt data capture to a very common software that is available everywhere in the enterprise. Finally, we have begun some experimentation with

SVG format. This is a vector graphical format that allows introducing lots of interactivity with the restitution. We think SVG is really promising in our problematic because it is entirely based on XML language. So it is really simple to handle this format and for example to translate a graphical representation in different way. It is very interesting to adapt restitution to symbolic representation that users are able to understand. After the global presentation of the Electronic Document Device that support the Diagnosis Guide, we propose to develop shortly some aspects in the specific use case of the Diagnostic Guide. 4.1 The document as a conception space

The approach that we perform in Electronic Document Computer System is based on the writing and the transformation of electronic documents. The use of electronic documents allows us to define a separation between data, formatting, structure and linking [8]: this is particularly suited to the technical contexts that are very structured fields. At first, we propose interface with weak formalized constraints and declarative data capture: each document corresponds to an Action Document as explained in [9]. That is a good way to avoid a formal data structure and too much environment constraint. In this use case of Diagnosis Guide, we conduct an experiment with a knowledge engineer that formalizes the results of his investigations, corresponding to capitalisation, with declarative data captures by using XML editor Oxygen (http://www.ogygenxml.com). Using XML language impose only to produce well formed document during the design of knowledge representation. The data capture is not only a transactional or procedural process that depends on data structure, but is nearer a constructivism process. The structuring and instances could grow together, until we obtain a stabilization of structuring that corresponds to a model. It is really different of a procedural data capture process where generally it is necessary to define knowledge model before. The structured content corresponds to knowledge representation. Then it is possible to extract from the document the canonical logical structure (Document Type Definition or XML Schema). This logical structure corresponds at least partially to knowledge model. Document is a design space, a design interface, as diagram in design software [6]. The data capture of data tagged information is not without problems. Thus, in the context of the Diagnostic guide, when the structure and the content grow up the knowledge engineer is confronted to lost in space problem [11]. It was difficult to locate itself

in the structured document. We introduce another XML editor (Morphon) to avoid at least partially this problem. It is a wysiwyg one because it allows to translate the tags by a graphical semantic (Figure 2). The definition of translation is based on CSS and in our context we have transformed the tags by frames in different colors.

chosen. Figure 3 and Figure 4 illustrate the result of the translation process for a textual and graphical representations.

Table 1. Correspondance between classification of information in textual or graphical representation system Information Class Entity Relation Identifier

Title

Description

Table + title

Similarity / Difference

Text form : bold, italic… Bi-dimensional space (document linearity) et Title size Writing the value

Order Figure 2. Morphon editor

Textual representation - Page break + title - title Hierarchy : Part, Section, Sub section…

Proportion

Graphical representation Node Edge Node label or Edge label Link to textual zone Color / Shape Bi-dimentional space Writing the value

Both editors that we have introduced provide types of restitutions fitted to some user and some activities (structuring document, data capture or update of content…) especially in the data capture process. Then we have worked on a mechanism that allow to translate one knowledge representation to different representation systems. 4.2 Formatting document to support meaning and interpreting Meaning and Interpreting of knowledge formalization depends firmly on structuring and formatting [5] [8]. So the choice of restitution modalities represents high responsibility in relationship with author activity or rather editor activity. To support that, we developed a supervised process that computes a translation from the knowledge representation (the formalization) to different semiotic or formatting systems (figure 1: textual or graphical). We take advantage of the data and formatting separation permitted by electronic documents. The restitution is computed thanks to a classification of information. This classification is defined according to picture theory [1] (especially the visual perception variables) and takes into consideration contextual parameters like task, activity or user specificities. One example of classification of informations and correspondance with representation systems is presented is Table1. As a result, we are able to compute restitution, influenced by efficiency notion [1], in order to propose to user a real object, an artefact, that helps analysis, cognition and activity [12] [13]. The translation process is conducted in order to lead the model semantic, with the help of the semiotic systems ability that we have

Figure 3. The Diagnosis Guide in HTML format (textual form) 4.3 A document that mediates a well-argued speech to support activity

In a Knowledge Management Context and especially when it supports a Decision-Making Process, one of the main objectives is to build a restitution that conveys a well-argued speech. This speech must be a consensus between an abstraction level that offers a global view of the problem and a detailed view that could respond to a very specific need. We can resolve, at least partially, this problem. In fact, it is possible to propose a panel of restitutions, with complex data filter process, computed according to strategies. For example, strategies must be influenced by the task, which the restitution must support [17][23]. Hence, each restitution could be seen as an Action Document, in relation with one point of view on data, and compliant with the activity that it must help.

allow browsing influenced by this point of view. Of course, it is possible to implement this without Xlink: one-way could be to use some unidirectional links associated with an algorithm that manages them to simulate the reality of a multidirectional link. The advantage of Xlink is that it provides a declarative data capture of links that are really flexible and directly readable and understandable by human and computer.

Figure 4. The Diagnosis Guide in SVG format (graphical form) In the design of the Diagnosis Guide, we have talk before about specific version of document focalized for one skill in the context of review process. That is represent a document computed by a very simple strategy. We think about the development of two another strategy. First, the Diagnosis Guide corresponds to a generalize representation of diagnosis on hydraulic power plants. We propose to personalize the document thanks to a representation of specificities of each hydraulic power plant. Second, in future work, it could be possible to propose a specific version of Diagnosis Guide for maintenance technician. Then the content could be organized in accordance with his patrol on hydraulic power plant. 4.4 Computing interactive and dynamic document

Moreover, following the evolution of hypermedia systems, we can highlight a relationship between this kind of system and KBS. Thus, hypermedia systems introduce a change in the interaction: it’s not longer a dialog driven by the system but interaction is nearer the user making his own reading path. Previously, we have presented various restitutions rather than a static form. Thus we plan to insert important interactivity in what we propose to users. This interactivity should make users become real actors (author or editor). To this end, we have started an experiments in two ways : • •

Using Xlink specification. Using AJAX (Asynchronous JavaScript and XML) development.

The Xlink standard allows linking relationships among more than two resources with complex transverse rules between them. It is an extension of the unidirectional link notion implemented in the case of HTML language. It provides an opportunity to compute restitution by different point of view and to

Figure 5. The first snapshot shows a digital document presented like a branching. For example, when we invoke the link labelled “item” (in red colour), the pointed chunk document (chapter in red colour) is embedded at the link position in the document (Second snapshot). We don’t assist to a move in the document (or remote document) that is limit the problem of lost in information space [11]. Moreover, the standard Xlink provides more than classical hyperlink behaviour. For example, the classical behaviour implies the opening of new windows with pointed document or the replacement of the document on screen by the pointed document: in the field of hypermedia and hypertext this situation is called a context move. By using Xlink, it’s really different because we have the possibility to perform classical behaviour already presented, but also we can augment interactively the restitution with the pointed document or only document chunk (Figure 5) without context move. This kind of function represents an interested way to enhance the interactivity, to place the user as an actor in restitution process to support his understanding, to

assist him in the making of reading way and to deal with the problem of lost in information space [11]. Ajax development is not a specific technology. It just corresponds to the association of technologies such as HTML, CSS, DOM, Javascript, XML and XSL. AJAX development introduces specific possibilities for web applications. Thus, classical web application need to submit request to a server to change the content of document loaded in the web browser. The difference with AJAX development is that allows to change a document in the web browser without the reload of the web page or the load of a new web page. In fact, a web page could be modify about the form or the content by asynchronous request with the web server. It is possible thanks to the use of the XMLHttpRequest that is now available on most of web browser such as Firefox, Internet Explorer and so on. This possibility is also due to the use of Javascript and the manipulation of Document Object Model (DOM) Tree directly in the web browser. Our interest for AJAX is particularly leaded by this possibility to modify a document without the need to reload all data from a server. With this mechanism, user can handle document interactively. At this time, it is really possible to transform the user in an actor in the restitution process. We can illustrate that with two examples. First, from an initial document, it is possible to offer the possibility to user to add information in relation with the information that is pointed with mouse pointer. We can talk of augmentation in information of the document. An example of this kind of mechanism is presented in Figure 6.

Figure 7. The first snapshot shows a document in the web browser that we define to a good integration in user activity. On the second snapshot a fragment on this document (highlighted) is rendered editable. Ajax libraries offer lot of another interactive mechanisms such as the possibility to hide part of document or to move part inside a document. We think that opens a wide exploratory space to improve interactively for facilitate appropriation and valorization of information by users. CONCLUSION

Figure 6. On this snapshot, we can see on the center a popup that is the result of AJAX asynchronous request Another example of AJAX using is to render a document directly editable. We think that is generally a mistake to propose a specific interface for edition process, generally based on formalize, that is different of the restitution that we develop with the efficiency notion [1]. The Figure 7 shows the possibility to transform part of document to an editable zone inside a web browser.

In this paper we have proposed an oriented document approach of Human Computer Interface in the context of Knowledge Base System. The document seems to a very flexible solution for the data capture process of knowledge and the restitution of knowledge representation. In the use case of Diagnosis Guide, we have applied our approach by the development of an Electronic Document Device. In this context, the document has shown his interest for the cooperation between people particularly knowledge engineer and experts. Moreover, the translation process that we propose seems to be a good opportunity to fit restitution to user and activity that could contribute to the sharing of knowledge and so to cooperation. One of the objectives with the document approach of Human Computer Interface is to transform the user to an actor in the restitution process. In fact, it is really difficult to predict needs of users, and generally

interfaces are more a way to get information in a standard format and not a way to handle it to support user activities. With the electronic structured document and thanks to the flexibility and the interactivity we try to propose a real solving space problem for the user that constitutes in a collaborative context an Action Document [9], a document as an interface.

[15] Schmidt, K., & Simone, C. (1996). Coordination

REFERENCES

[16] Gilles Falquet, Luka Nerima, Jacques Guyot,

[1] Bertin, J., “Sémiologie graphique - Les diagrammes - Les réseaux - Les cartes”, Les réimpressions des Éditions de l'École des Hautes Études en Sciences Sociales, 1973. [2] Bourguin, G., Derycke, A., and Tarby. J.C., “Systèmes interactifs en co évolution réflexions sur les apports de la théorie de l’activité au support des pratiques collectives distribuées”. RIHM, 6(1), 2005. [3] Dieng-Kuntz, R., Corby, O., Gandon, F., Giboin, A., Golebiowska, J., Matta, N., and Ribière. M., “Méthodes et outils pour la gestion des connaissances : Une approche pluridisciplinaire du Knowledge Management (2ème édition)”. Dunod, 2001. [4] Dourish, P., Where the action is – The Foundations of Embodied Interaction. MIT Press, 1999. [5] Eco, U., Le Signe, 1988. [6] Morand. B., “Logique de la conception - Figures de sémiotique générale d’après Charles S. Pierce”. L’Harmattan, 2004. [7] Ouni, A. and Dudezert. A., “Etat de l’art des approches de définition du système de gestion des connaissances (sgc)”. In Systèmes d’information : perspectives critiques, INT Evry, Mai 2004. Association Information et Management. [8] Pédauque. R.T., “Document : forme, signe, médium, les reformulations du numérique”. Technical report, STIC-CNRS, 2003. [9] Zacklad, M., “Processus de documentation dans les documents pour l’action (dopa) : statut des annotations et technologies de la coopération associées”. In Le numérique : Impact sur le cycle de vie du document pour une analyse interdisciplinaire, Montréal, 2004. EBSI. [10] Zacklad, M., Legay, N., and Blanquet, J., “La modélisation conceptuelle des ihm avec pss”, In actes du 5ème Colloque Ergo-IA, pp101–115, Octobre 1996. [11] Cox, R., Brna, P., “Supporting the Use of External Representations in Problem Solving : the need for flexible learning environnements”, In Journal of Artificial Intelligence in Education, 239-302, 1995. [12] Norman, D.A., “Cognitive artefacts”, Designing interaction, 1991. [13] Norman, D.A. “Affordances, Conventions and Design”, Interactions, 38-42, Vol 3, N° 1, 1999. [14] BRASSAC, C. and GREGORI, N., “Co-construction de sens en situation de conception

d'un outil didactique”, Studia Romanic Posnaniensia, 55-66, 2000 mechanisms: Towards a conceptual

foundation of

CSCW systems design. CSCW Journal, 5, 2-3, 155-200. Christine Vanoirbeek, Yassine A. Rekik (1999). Des documents virtuels pour lire les bases de données. IHM 99 - atelier Documents Virtuels personalisés. [17] Béguin, P. and Rabardel, P. (2000). Concevoir pour

les

activités

instrumentées.

Interaction

Homme-système, perspectives et recherches psycho ergonomiques, Revue IA, 35-54. [18] Kovacs, Brice and Gaunet, Florence and Briffault, Xavier (2004). Les techniques d'analyse de l'activité pour l'IHM. Hermès Science. [19] Dufresne, Aude (2001). Conception d'une interface adaptée aux activités de l'éducation à distance – ExploraGraph. Sciences et Techniques Educatives. [20] Boullier, Dominique (2001). Les choix techniques sont des choix pédagogiques. Sciences et Techniques Educatives. [21] Parfouru, S., Grassaud, A., Mahé, S. & Zacklad, M. (2006). Document pour l’Action comme media pour la Gestion des Connaissances. 9ème Colloque Internationale

sur

le

Document

Electronique,

199-215. [22] G. E. Krasner et S. T. Pope. A cookbook for using the

Model-View-

.

Controller

user

interface

paradigm in Smalltalk-80. Journal of Objet Oriented Programming, pages 26–49, 1988. [23] Kaptelinin, Victor (2003). Learning with artefacts : integrating technologies into activities. Interacting with Computer, Vol. 15, 831-836. [24] Hatchuel A., and Weil, B. (1992). L’expert et le système,

suivi

de

Quatre

histoires

de

systèmes-experts. Economica. [25] Ermine, J.L. (2000). La gestion des connaissances un levier stratégique pour les entreprises. Actes de IC 2000. [26] Michel, Grunstein and JP, Barthès (1996). AN Industrial View of the Process of Capitalizing Knowledge. Acte de la conférence ISMICK. [27] Haïk, Philippe and Mahé, Sylvain and Ricard, Benoit (2002). Knowledge engineering as a support

for decision making

in plant operation

and

maintenance. Pacific Rim Knowledge Acquition Workshop, Tokyo, Japan.

[28] Casabianca M. . (2005) Extreme Programming precis et concis, O’REILLY.