AN ORIGINAL MODEL TO ORGANIZE KNOW-HOW IN A BENCHMARKING CONTEXT

MAHE S. (*), RIEU C. (**), BEAUCHENE D. (***)
LLP / CESALP
Laboratoire de Logiciels pour la Productique
41, av. de la Plaine, BP 806
74016 ANNECY CEDEX
FRANCE
mahe@esia.univ-savoie.fr, rieu@esia.univ-savoie.fr, beauchene@univ-savoie.fr

(*) preparing a Phd in Computer Science LLP/CESALP
(**) Computer Science lecturer at the university of Savoy- France
(***) Phd in Computer Science LLP/CESALP

Abstract: This paper presents an approach to capitalize know-how through a benchmarking project. We explain firstly why a quality management project has led us to study corporate memory and present the synergy between these two domains. We find in a second part a theorical analysis of know-how pointing its dynamic aspects. The worth of our approach consists in considering both explicit and tacit knowledge and in structuring know-how by attaching it to the organization of the enterprise or of the network of enterprises (in the benchmarking context). To identify and capture know-how to transfer we intend to use groupware tools for building the benchmarking knowledge based-system.

INTRODUCTION

Today, competition and flexibility demands lead companies to find improvements by applying new strategies and structures. The necessary adaptation happens inevitably through innovation, that is to say through an efficient and dynamic management of the change. Companies have to face the same quality challenge; the referential (norms) strive to be universal, however each company has its own organization, its history and its culture.

Face to these challenges, knowledge is now recognized as a major and strategic subject for enterprises in an economical context of world-wide competition. Managing knowledge is not easy since corporate knowledge is spread amongst diverse sources throughout an organization, its environment and manifests itself in related knowledge networks..

We present in this article a model to organize knowledge and to assist enterprises in the difficult task of know-how capitalization.

We firstly introduce the research project that led us to focus on the know-how capitalization in the enterprise. It concerns a Franco-Swiss inter-regional project of benchmarking between companies of different economic sectors and located on both sides of the border.

One of the purposes of this project consists in initializing a benchmarking knowledge base that memorizes the various know-how used during this process in order to make this know-how reusable during future benchmarking. The aim of a benchmarking process is of course to share know-how with partners, but it is also a problem of studying how knowledge is shared and transferred inside the enterprise. But what do we call "know-how"? Is it knowledge, experience, expertise, etc...? Many works have already define these terms in different research area. We devote our second part to the presentation of the different types of know-how that we consider in our approach and their characteristics.

Then, in the third part, we situate our work among different methods to capitalize know-how. We identify and model know-how in companies by associating it directly and in a dynamic manner to the enterprise organization. We use in our approach the Olympios model developed by the LLP-CESALP.

Finally, in the fourth part, we introduce an environment to help know-how exchanges. We analyse the existing links that exist between the different kinds of information we retained. We describe two specification levels for this knowledge base: an internal level and a global sharable level that would allow companies to share know-how on the World Wide Web of Internet. For the " internal level ", techniques for the knowledge acquisition determine the knowledge structure. We introduce briefly some groupware techniques as useful tools to capture knowledge not easy to specify. We intend to go further in this research area.

1. A BENCHMARKING CONTEXT

1.1 A need for continuous improvment

For the time being, it is essential for small business to be well ranked in a moving world market. In order to be in accurate shape for this challenge, they turn towards a new certification approach, either imposed by their clients, or initiated within the context of a service improvement policy.

They have then to master all necessary conditions for the quality building and maintenance expected by the customer and that means a large rigor all along their production, the installation of a better adapted organization, and the respect of procedures in all enterprise areas (Lamprecht, 1994). To undertake this continuous improvement approach, small business lack experience. Indeed, most of the time, an enterprise feels difficulties to identify its strong points and its weak points by evaluating with an accurate manner its current performance level.

Among tools that can help an enterprise to overcome these difficulties, the benchmarking is being established as one of the most efficient to allow the enterprise to enter efficiently in a continuous improvement approach (Pôle Productique 1995). Thanks to American enterprises that are increasingly numerous using it in an intensive manner, the benchmarking has become a complete management tool, well defined and with a well established process. It is therefore natural to let French and Swiss enterprises take benefit from it, whose major current concern is to continually improve whether this is to survive or to become the best in what they make.

1.2 A Franco-Swiss benchmarking project

The work that we present is a part of a global project of benchmarking between French and Swiss companies. The main aim of this project is to develop computer-aided tools helping small business to improve their performance. This global project has three main objectives:

* The first aim of this project is to realize a benchmarking gathering French and Swiss enterprises in order to improve their performances.

Benchmarking is a continuous evaluation process for products, services and methods. It leads each enterprise to compare itself with serious competitors or leaders (Camp, 1992; Karlöf & partners, 1994). Either the enterprise adopts their practices, or it adapts them with the aim of improving its performance.

The idea of benchmarking is not brand new and plays an increasingly important role in improvement approach used by Swiss as well as French enterprises. However, most of benchmarking actions are limited to a local and homogeneous partnership (same sector and same geographical area) (functional benchmarking). But companies in different areas could also have a great interest in sharing their know-how (generic benchmarking).

* The implementation of the benchmarking that we propose needs to rely on a certain number of methodological supports and tools. Some of them exist, others will be developed in the context of this project. Our next objective is therefore to lead to the realization of a toolbox supporting the whole steps of a benchmarking process.

* Finally, our third objective is to generalize previous works by initiating a knowledge base capitalizing results of other benchmarkings that will be undertaken. But we don't only want to storage the more significant indicators, but we need to memorize all the use context of these indicators, that is to say practices and methods to obtain them. It is really the enterprise know-how that must be capitalized in order to reuse it inside the enterprise or to share it during future benchmarking.

Among the products of this project, we propose two types of software :

* a software for an internal evaluation of the enterprise (by characterization and grading of located dysfunctions), which compares the enterprise performance level with standard values.

* a software and a knowledge base which guide step by step a benchmarking process (external performance evaluation).

These tools are completed by a development methodology to realize the different phases of a benchmarking.

The project has three types of partners :

* a research laboratory, " Laboratoire de Logiciels pour la Productique " in Annecy (LLP/CESALP) of Savoy University

* French and Swiss transfer centers

* fifteen enterprises located on both sides of the Franco-Swiss border, representing experimental platforms of the project. These enterprises have quite different sizes and are on various sectors (mechanics, electronics, food processing, ...).

We can find in (Mahé, Maire & Rieu, 1996) a more detailed presentation of this project, especially concerning the internal diagnosis phase and its results. This article develops the know-how capitalization through the benchmarking process.

1.3 From benchmarking to know-how capitalization

We have just shown how this project of benchmarking leads us to study the theme of the know-how capitalization. In the framework of this project it concerns especially exchanges of very specific know-how with external partners; but at the same time it is as important for each enterprise to valorize all know-how transfers inside the enterprise.

" In the USA., in a steelworks the adjustment of the train of mill has become impossible since the premature departure of some old workers. In a chemical unit, the beginning of a complex installation is in wait so that the superintendent acts personally on some levers which are known by him alone after 25 years in company. A workshop of automobile part manufacture is modernized. The automation is elaborate to the extreme, outputs expected in quality of the new produced parts, are very difficult to reach. They are even inferior to preceding performances. " (Loubet, 1992)

The common denominator of these three examples is: one does not know how to do, one no longer knows how to do. It is really the know-how that we have not succeeded to acquire or to keep. This is an element of knowledge that conditions the efficiency of an enterprise until its durability. According to D. Loubet, we manage with care the tool of production, and we manage badly workforce. Even with a consequent workforce management service, that is serious and that tries to be objective, in practice the system " appears terribly subjective and the vocabulary used to judge individual performance is dramatically poor. A semantic analysis would show that only some words are used... "

" Tool of Production, Workforce, would it lack a dimension ? Yes, to capture, formalize, perpetuate the know-how " that is one part of the enterprise patrimony.

D. Loubet presents us in his introduction (1992) one of the main two reasons for the growing interest that enterprises currently develop for know-how management.

These main two reasons (Mayère, 1995) are on the one hand the loss of know-how due to movements of employees (retirement, promotion, departure...) and, on the other hand, the quality management that needs to write practices, to homogenize them and to optimize them in an improvement optic.

In the case of the loss of knowledge, it is only after a certain lapse of time, that the interest for the capitalization of the know-how emerges in a cumulative and reactive manner. Thus in a logic of risk management, we invest gladly in know-how management to avoid the occurrence of malfunctioning and we wait the return on investment that should have to follow.

The second reason of this interest is also situated in a process that is economically justified for the enterprise and that consists in doing an investment - in quality management - for making its process more reliable by formalizing its know-how in procedures, waiting also the return on investment that we can rightfully expect, if the quality approach is not only suffered.

Then, why do these preoccupations arrive only now? It is the emergence of a " new industrial model " (Mayère, 1995) that imposes us these new preoccupations. From mass production, mass organization and mass marketing, we arrive to flexible production, to niche marketing and to networked organizations (Webber, 1993). The revolution in communication and information technologies makes knowledge the new competitive resource placing individuals in the foreground. These individuals possess knowledge and can learn some new one from their personal experience, their colleagues, their clients, suppliers and economic partners. In order to do that, the organization, specially the top management have to construct the environment that allows the individuals to create knowledge. It is notably by favoring the communication that one can help individuals in such activities. It is the reason why we will be going to seek replies in the groupware systems.

2. DEFINITION OF THE KNOW-HOW

In this part, we specify know-how typologies and characteristics that we have decided to retain. That will allow us to put in place the existing terminology of close notions as: know-how, know, knowledge, expertise, experience...

2.1. Tacit or explicit know-how

The stemming distinction of Polanyi (1969) allows to distinguish two types of knowledge : on the one hand the explicit knowledge (formalized in books or by words...) and on the other hand the tacit knowledge (non-formalized, fruit of the experience). This distinction has been made for a long time, under various forms that can be neared.

Tacit and explicit knowledge characteristics

Explicit knowledge corresponds to formalized knowledge, notably knowledge of books, and more generally, that is transported by the language (written or spoken). With a known syntax and a known semantics, the explicit knowledge can be transmitted from a person to another by the language without loss of a part of this knowledge. Transfers of this type of knowledge are relatively easy, since they are acquired by the language, they are naturally expressed by the language.

But, we can identify another characteristic of explicit knowledge, that is the fact that it has a general character, distanced from practice. This is tacit knowledge that is close to the practice.

The tacit knowledge (or implicit) corresponds to non formalized knowledge, that Nonaka (1994) defines in the following manner: "the tacit knowledge includes both cognitive elements, diagrams, beliefs, mental models... defining our vision of the world and of technical elements corresponding to a know-how anchored in specific contexts of action". As far as the tacit knowledge is not acquired by the speech but by imitation or experimentation process, it is impossible or very difficult to translate in speech (it necessitates the formalization process detailed in 2.2.). In this meaning, Polanyi (1969) asserts us: " we can know more than we can tell ". Indeed, this tacit knowledge really lives in each of us, but we have often a lot of difficulties to express it.

We distinguish two types of tacit knowledge:

* context knowledge,

it corresponds to a set of norms and implicit values more or less shared (culture, behavior...),

* practice knowledge (know-how),

that Kogut & Zender (1992) define as know-how: " particular expertise that allows to realize something easily and efficiently and that is acquired by experience ". " The essential of the " modus operandi " that defines the mastery of the practice is transmitted by the practice, in the state of practice, without accessing the level of the speech " (Bourdieu, 1980). The practice knowledge, or procedural knowledge (in the Anderson ACT model), corresponds to a process, that, contrarily to the declarative knowledge (ACT but not exactly equal to explicit), describes a state. This knowledge is acquired in practice, as automatic actions (we can speak about " stimuli ") triggered in a very precise and discrete (in the mathematical sense) context. That's why we associate to this knowledge the context of action that conditions it strongly.

Implied individual behaviors are stocked in their procedural memory under a relatively automatic and inarticulate form, which explains the difficulty to communicate it by the language.


Explicit knowledge

Tacit knowledge
Decomposition

context knowledge

practice knowledge (know-how)

Practice
far from practice, theoretical
deeply rooted in practice
Context
not precise, general concern
needs a very precise context of execution
Exchange
language spoken or written
shared experience,

observation, imitation, practice

Example
procedure
knack

Figure 1. Organization knowledge according to the tacit / explicit typology

2.2. Dynamics: knowledge nature changes

The enterprise wealth and culture are henceforth less measured in the term of an accumulated capital than in capacity of circulation. J.L. Lemoigne speaks about a " permanent process by which doing transforms itself into know-how that transforms itself into doing " (1993). That's why it is necessary to apprehend the knowledge in the enterprise in a dynamic manner.

Nonaka (1994) has studied the dynamics of knowledge with regards to the tacit / explicit distinction. According to him, the creation of the knowledge within the organization implies the transmission of knowledge between individuals and a change of nature for this knowledge; that is to say four possible evolutions (from one to same or to the other):

* From Tacit to Tacit : the socialization

By strong interaction the tacit knowledge of a person can become the knowledge of other persons, without needing to use the language. It is notably the case for the apprenticeship of a knack : it is not acquired by the apprentice with the help of a language but with observation, imitation and practice. It is in the practice shared by several persons, called socialization by Nonaka, that creations and exchanges of tacit knowledge occur.

* From Explicit to Explicit : the combination

With a common language and communication mechanisms, people move diverse knowledge elements closer, combine them; and obtain thus new knowledge (by induction or deduction...). This is the easiest to mention mecanism, but it's only one of the four. The new knowledge created within this combination process is theoretical, and cannot be used in practice at this stage (it needs the 4deg. process).

* From Tacit to Explicit: the Externalization (formalization)

Efficient practices are clarified in a formalized speech. In order to do that, we have often recourse to the metaphor and to the reasoning by analogy (for landing the shared concept absence). This conceptualization is an operation heavy to implement. Indeed, practices are experimented without reaching the level of concepts, but to make them explicit and to communicate them by the language, it is necessary to bring them to the level of concepts, that is to say to make relationships between them and generalizations. This operation has been identified by Piaget for child in the passage from the sensorimotor intelligence to the representative intelligence.

* From Explicit to Tacit: the Interiorisation

By repetition, we root the explicit knowledge in sequences that can reach the reflex level, by adapting the explicit diagram to specific conditions of the execution. Starting from knowledge formalized in the language, we arrive with this interiorization process to apply this knowledge in practice, that is to say to create some tacit knowledge from explicit knowledge.

This classification allows us to identify several manners to create the knowledge within the organization. Nonaka explain that the socialization has been approached in the context of organizational culture, combination is largely studied by the data-processing (processing of formalized information, or data), internalization has been covered in the framework of the organizational learning, but the last dimension, externalization (tacit formalization), has not been yet well developed. In order to do that, we can use techniques implemented inside the groupware systems that allow to capture information to identify elements to formalize. We can also use mecanisms implemented in tools for meeting that allow to capture parts of formalization of these elements (externalisation).

Figure 2. Dynamics of knowledge creation in the enterprise

Also, for a system the aim of which is to help to capitalize enterprise knowledge and to create new knowledge (strategic resource), we must take into account not only the two static knowledge dimensions (2.1.), but also the four dynamic knowledge dimensions (2.2.).

2.3. Collective or individual know-how:

We just precise here that knowledge of a collective work group is not only the sum of the individual know-how of persons who compose the group, and we will see later how our approach takes into account individual know-how as well as collective know-how.

We can note that collective and individual levels exist also for the practice knowledge : the collective practice corresponds for Cohen & Bacdayan (1994) to repetitious sequences of acquired behaviors implying several actors in communication within a community of practice.

3. SEVERAL APPROACHES FOR KNOW-HOW CAPITALIZATION

We can distinguish several approaches according to various disciplines for know-how management. We introduce in this part only a few of them in order to place our work among existing methods and to show what is our " added worth ".

3.1. Competence Management

First, the world of Human Resources Management has leaned on this problem for several reasons. Previously as a criterion for person evaluation, and later for career management, it concerns now, with the new organization forms, the capability to find who would be the most efficient (person or group of persons) in the context of a new project. Competence is the necessary criterion to realize this selection. One can build a catalogue of all competence of individuals in the enterprise by identifying and storing competence of each worker. It is then possible to consult this catalogue to find, inside the enterprise, who are individuals that possess the most (or the totality) of the competencies researched.

A system developed with the concept of knowledge tree (Authier & Levy, 1992) allows to reply to these issues of human resources by providing a way to organize competence in a tree following several criteria as the main trade, the strategic importance for the enterprise... With this software tool (Trivium, 1996), we can select a set of competencies and to find the persons the closest to this profile. This environment allows to cover the knowledge tree to see who holds such knowledge and to question the person on his / her knowledge or know-how through this environment, since this system has some groupware facilities for communication.

3.2. Expertise Capitalization

Expert systems and knowledge based systems allow also to reply partly to this problem of know-how capitalization. Their optic is to group knowledge of a very precise domain, limited to one or few experts. This approach is very heavy since it necessitates to formalize all a part of knowledge of experts (relevant for the objectives of the expert system) in view to reproduce their diagnosis automatically. And because of tacit knowledgethat can't be formalized, the obtained result is often far from the real representation of the experts know-how.This operation is therefore adapted to limited operations in space and in time.

3.3. Quality approach

As contrast with the last approach, the quality management implies a global approach of the enterprise. It aims to write all processes implemented in the enterprise in procedures to make practices more reliable. It is also a process of knowledge formalization limited at the human language without the precision and completeness levels that an expert system needs. But quality approach does not always take into account the whole practice, it is often perceived too tedious and useless, without evoking aspects of confidentiality that bring also brakes to the process.

3.4. Our approach: know-how in the enterprise organization

Face to the limits of the previous approaches, we propose to build a system to assist know-how capitalization that takes account of tacit knowledge and where know-how is directly " attached " to the organization of the enterprise. Know-how is thus directly linked to its context of use.

3.4.1. Enterprise organization model

We have chosen a representation of the organization leaning on the OLYMPIOS model developed by the LLP-CESALP since 1989 (Beauchêne & Maire, 1995). We summerize here in a very simplified manner the characteristics of this model. In OLYMPIOS, the organization of the enterprise is seen through the " customer - supplier " relationships that exist between the different actors. These relationships are established according to a purpose to reach (even if this purpose is not always clarified). The exchange relationship has therefore to be defined from the objective to reach. This objective is translated by customer through the expression of a need and transmitted to the supplier. This last one is responsible of the supply of a resource that is able to satisfy this need.

Finally, so as to ensure following of the performance of the enterprise, it is indispensable to evaluate the adequacy of the supply with the objective. The OLYMPIOS model names satisfaction level the result of this evaluation.

Thus, the atom of organization in OLYMPIOS is the " Customer Supplier Information System " (CSIS). A simplified representation is given in figure 3.

Figure 3. Atom of organisation in OLYMPICS model: the Consumer-Supplier Information System

We can find more details on this model and its use in the context of know-how capitalization in (Beauchêne, Mahé & Rieu, 1996). Know-how of each actor can therefore be situated regarding his role in the functioning of the CSIS (customer or supplier).

3.4.2. Inter-entreprises model

The Olympios model has been developped initially to represent exchanges relationships inside the enterprise in a context of quality management; but we can easily extend the use of this model in the case of exchanges between partners in a benchmarking context. Indeed , in the atom of representation called " Customer-Supplier Information System " (CSIS), the custumer may be an enterprise asking the supplier (another enterprise) to share its know-how.

Interconnections between severals companies will be modeled by a network of CSIS. We rely on this global representation to identify and capture the know-how to transfer, with its contextual aspect.

4. STRUCTURING THE KNOWLEDGE BASE SYSTEM

4.1. A global approach : groupware aspect

Groupware techniques allow us to reply to many specifications previously quoted for our environment for know-how capitalization and benchmarking. Indeed, to adapt to new demands of organizations in term of working in groups, in term of information and knowledge circulation, techniques implemented in groupware systems give already some elements of consequent reply. It is in this direction that it is necessary to advance, but from now on, we can specify mechanisms that seem already to be adapted to our problematical.

4.1.1. Facilitating communication and putting people in contact

Facilitating the communication and exchanges of knowledge in the enterprise allows to favor the creation of knowledge in the enterprise (Nonaka, 1994). In order to do that, some groupware software tools exist to communicate (messaging, forum) or to navigate in a common representation, to find an information or a speaker according to different natures of criteria or according to an element that one researches. Thus research engines implemented in web browsers, for example, can help us to find a piece of information inside the complex representation. There are also some mechanisms such as coordination with subscription / notification (Gruber, Tenenbaum & Weber,1992; McGuire, Kuokka, Weber, Tenenbaum, Gruber & Olsen, 1993; Gronbaek, Hem, Maadsen & Sloth, 1994; Mahé, 1994) that can help us in an active manner : from explicit or automatic subscription, the system will do some actions as giving a piece of information, or for example, telling someone that another person wants to know some details about something that the first one knows.

Therefore, individuals in the organization can be helped in contacting each other in order to exchange various natures of knowledge. Especially, this help for relationship can be a starting point to the socialization between two individuals, and consequently, these mechanisms can favor it.

4.1.2. Capture of non structured information

During non-structured exchanges as through messaging systems, some information can be captured automatically and hypotheses can be made on the manner to link them to a precise place in our representation of the organization. These elements have a rough nature and they are integrated in the system with this status. However, this technique gives some starting points to formalize operations, after a first step of identification of the knowledge to formalize. It is also that information that will be used for notification / (automatic) subscription techniques.

4.1.3. Capture of semi-structured information

Many works have been done in the framework of computer-aided system for meetings, not only to allow meetings from a distance, but also automatic reviews, or in a lesser extent to keep traces of interactions that occurred during the meeting. It is notably the case of systems whose objective is to capture the reasoning that leads to the decision-making, in order to reuse this reasoning later. We can find this type of mechanisms in such tools as The coordinator ,Lux de Greef, Bomarius & Steiner, 1993), gIBIS (Lee, 1990) and SIBYL (Lee, 1990). These systems use semi-structured representations to model the interactions during the meeting. A problem has been identified by Conklin (1996) about the means to access captured knowledge in these systems in order to reuse it : it lacks a link to an effective structure. The bond with the representation of the organization allows us to reply to the necessity of having a link to structure. These links being constructed from precise points of the organization, that allows us to consider the context of realization and not only the results. This responds also to a suitable critic of Conklin on current systems to consider results (product, document...) rather than process that lead to these result (western culture, centered on results in exit).

4.2. Toward a common representation

The interest of a common representation is to empower sharing information within the organization and also for our problem of capitalization to empower knowledge exchange and creation.

4.2.1. Structure of the common representation

The structure of the common representation integrates several types of information with links between these informations. We have retained four types of information that are :

* representation of the Organization (O),

* representation of Know-How (KH),

* representation of Captured Elements For Identification (CEFI), (cf 4.1.2.)

* representation of Elements Captured during meetings (EC) (cf 4.1.3.).

We can note the difference between EC and CEFI elements according to groupware mechanisms that bring them. CEFI is an uncertain information captured automatically during communications between persons while EC is semi-structured information capture within meeting tools such as those cited in 4.1.3.

Here is an example of the representation of the four structures and the links between them.

Figure 4. Example of common representation.

These four elements are directly linked to persons or groups of persons (collective know-how) in the enterprise organization. Know-how is actuated by individuals and we can ask these persons for more information about their know-how. It is also interesting to know who has participated in a meeting as far as we capture EC during it. There would be some great interests to build these links but they will not be detailed here.

In the context of our model, it is interesting to specify links between each of these four elements (O, KH, CEFI and EC). Several kinds of links (and semantics) exist according to the type of the linked information.

4.2.2. Why building links between the four representations

Classical approaches retain most of the time only one representation and leave out the others (that would lead for our environment to memorize CEFI, KH and EC with each concerned O-elements, and thus destroying inter-CEFI, inter-EC and inter-KH organizations. We think that it is important to keep separated these four representations and to create links between them.

* KH - O links

An element of know-how is linked to a point in the organization.

* KH O

We can tell where in the organisation this know-how takes place. This point in the organization gives a large part of the context of know-how.

* KH O

We can tell which elements of know-how an element of the organization needs. This is important for the exchange of efficient practices (in the context of the benchmarking process) to be able to identify the elements of know-how that a process (or a part of a process) implies.

With the two oriented links we can construct links between elements of know-how (KH1 O KH2).

* CEFI - O links

CEFI are elements captured to identify possible existence of know-how (that should be captured later). In various points in the organization, we can capture this types of information and link them to these respective points in the organization representation.

* CEFI O

* Starting with a CEFI, it is possible to reach a point in the organization. This could be very interesting for navigating in the common representation. For example, if we want to know where a tool is used in the organization, we can use a lexical analysis to find in the CEFI plan which elements mention this tool. Then, from these CEFI we reach points in the organization and so, we can know where this tool is used in the enterprise organization.

* CEFI O

In a point of the organization, we can see the possible know-how to capture. In this way, the possible know-how is identified and it just remains to formalize it if necessary.

* EC - O links

These links are roughly the same as CEFI - O links; they are also used to identify know-how but moreover they help to formalize tasks.

* KH - CEFI links

* KH CEFI

A use of this link is to identify information related to an element of know-how.

* KH CEFI

With this link, we can reach an element of know-how. In addition to navigation and data mining techniques this can help us with drawing links between elements of know-how (KH1 CEFI KH2).

* EC - CEFI links

These links between captured elements should not be implemented directly. This is through links EC - O & O - CEFI, or EC - KH & KH - CEFI that these links are virtually implemented. The only interest for implementing this type of link would be to capture CEFI during meetings when information can't be included in EC (semi-structured) and thus memorized in CEFI and linked to one EC element.

* KH - EC links

* KH EC

The same interest as KH CEFI but for semi-structured information.

* KH EC

The same use as KH CEFI but for semi-structured information.

Through these elements and the links between them, we have just shown how we can represent a lot of information for structuring know-how in the organization representation, but also to give a multi-faceted context to know-how and to help the construction (identification and formalization) of know-how that has no use without its context.

4.2.3. Navigation in the common representation

To navigate in this common representation, it is necessary that individuals have effective tools to help them. Indeed there are several representations and several types of links between them. The difficulty has been identified and processed in the framework of hypertext systems, confronted with the same difficulties of navigation in a complex structure created by the multiple links that exist between documents in these systems. Management of organization evolving add another dimension to this complexity of links and leads us to implement versioning mechanisms to keep the organization history (in the framework of our organization representation).

4.3. Two logical levels for our benchmarking knowledge base

The operation of benchmarking presented into the paragraph 1.1. has to end to the construction of a knowledge base capitalizing the realized experience and reusable from two levels:

* on the one hand, in an internal manner for each enterprise; it will concern to preserve the detail of all practice occurrence in this benchmarking context in a knowledge base for the enterprise. This knowledge base will be able to be enriched with multiple know-how implemented in others contexts than benchmarking.

* on the other hand, in a global and external manner; it concerns then to construct a knowledge base shared by the different partners of the benchmarking process, in which are preserved the "best practice" stake at disposal of everyone. The selection of such know-how will have nevertheless to respect rules of confidentiality imposed by each enterprise. In this knowledge base we don't need to formalize know-how in details but only identify it. This knowledge base will be able to be used and enriched during future benchmarking operations between new partners. We have chosen to share this benchmarking knowledge base through World Wide Web, as, to our minds, this communication medium must be reckoned with nowadays, even if only few French small enterprises are already connected.

Each enterprise has a local environnement enabling to capitalize its know-how. It communicates with other enterprises through a shared environment as follows:

Figure 5. Inter-Enterprises Common Environment

We have then to define how knowledge can go from one level to the other. That has to be covered from two extents. First, from benchmarking point of view, there are exchange of know-how between enterprises that sign a contract (with some limits). So, in our environment, the exchange from intra level to inter level has to be validated by someone inside the concerned enterprise, as this could be confidential information, for example.

The second extent is the way to make part of representations from several enterprises cohabit in the inter level representation. That leads to examine some aspects of the domain that covers knowledge exchange between several KB (Knowledge Bases) and eventually mechanisms for achieving consensus ( Mahé, 1994).

We have now stated for the same environment for the two levels. Indeed, this is exactly the same inside the enterprise (exchange of know-how between people inside the enterprise) and outside (exchange of know-how between people from several enterprises) except the problem of contract that is less difficult for inter level than for intra level (but can also exist in inter level : when, inside one enterprise, someone asks to another to teach him how to do something, the so become " teacher " (in addition of his used work) may expect a return from his " learner ").

We intend to go further in this research area.

CONCLUSION AND PERSPECTIVES

In this paper we have presented a system to help enterprises to capitalize their know-how in a benchmarking context. The worth of our work consists in putting employees at the heart of the matter. The members of the enterprise will navigate through a common representation of the organization inside the entreprise. This model can be extended in the context of benchmarking to an inter-companies common environment, that proves the synergy between know-how capitalization and quality management.

Another original aspect of our approach consists in considering both explicit and tacit knowledge and in structuring know-how by attaching it to the organization of the enterprise or of the network of companies (in the benchmarking context).

To identify and capture know-how to transfer we intend to use groupware tools for building the benchmarking knowledge based-system.

We have explained why a quality management project has led us to study corporate memory and the synergy between these two domains. We detail also a theoretical presentation of know-how pointing its dynamic aspects.

Our approach is relatively theoretical except for the internal diagnosis approach that has ever given some practical results. This is only a temporary state, since we are entering a large experimentation phase in parallel of the continuation of the research parts. The new software corresponding to OLYMPIOS model applied for diagnosis adapted for benchmarking (described in Mahé, Maire & Rieu, 96) will be used in the next few days in a 300 people industrial business for the both context of know-how capitalization (identification and organization representation) and benchmarking (internal diagnosis) in order to validate choices that have ever been taken, and to refine the model. Later, this will be practiced in the fifteen other enterprises that participate to our benchmarking project.

BIBLIOGRAPHY

Authier M., Levy P., Les arbres de connaissance, Trivium, Juin 1992.

Beauchêne D., Mahé S., Rieu C., Enterprise Know-How Capitalization and Benchmarking with an enterprise organization model, Fourth International Symposium on the Management of Industrial and Corporate Knowledge, ISMICK'96, Rotterdam, The Netherlands, October, 1996.

Bourdieu P., Le sens pratique, Editions de minuit, 1980.

Camp R.C., Le Benchmarking, pour atteindre l'exellence et dépasser vos concurrents, Les Editions d'Organisation, 1992.

Cohen M., Bacdayan P., Organizational routines are stored as procedural memory : evidence from a laboratory study, Organization Science, vol. 5, ndeg. 4, p.554-568, novembre 1994.

Conklin Jeff, Designing Organisational Memory : Preserving Intellectual Assets in a Knowledge Economy, Corporate Memory Systems Inc., Austin, Texas, 1996.

Gronbaek K., Hem J.A., Maadsen O.L., Sloth L., Systems : A Dexter-Based Architecture, Cooperative Hypermedia, Communicatioons of The ACM, Vol. 37, No 2 pp 65-74, February 1994.

Gruber T.R., Tenenbaum J.M., Weber J.C., Toward a Knowledge Medium for Collaborative Product Development, Artificial Intelligence in Design' 92, Proceedings, Kluwer Academic Publishers, 1992.

Karlöf & partners, Pratiquer le Benchmarking, Les Editions d'Organisation, 1994.

Kogut B., Zander U., Knowledge of the firm, combinative capabilities and the replication of technology, Organisation Science, vol. 3, ndeg. 3, p. 383-397, août 1992.

Lamprecht J.L., Se préparer à la certification, Traduit par l'AFNOR, AFNOR Gestion Qualité, 1994.

Lee J., SIBYL : A Tool for Managing Group Decision Rationale, CSCW 90 Proceedings p 157-168, ACM Press, October 1990.

Lemoigne J.L., Sur l'Ingénierie de la connaissance organisationnelle, Note de recherche ndeg.93-02, Grasce, Janvier 1993

Loubet D., Le management des savoir-faire de l'entreprise, les Editions d'Organisation, 1992.

Lux A., de Greef P., Bomarius F., Steiner D., A Generic Framework for Human Computer Cooperation, ICICIS 93, International Conference on Intelligent and Cooperative Information Systems, Proceedings, IEEE Computer Society Press, 1993.

Mahé S., Des collecticiels ^ la construction concourante de bases de connaissances, DEA Système d'Information, Université Joseph Fourier, Grenoble, 1994.

Mahé S., Maire J.L., Rieu C., To evaluate and to improve the performances of the enterprise : a Franco-Swiss benchmarking project, , Modeling Techniques, BPR and Benchmarking International Workshop, Bordeaux, France, April 1996.

Maire J.L., OLYMPIOS un modèle de conception du système d'information d'une entreprise manufacturière - application à l'audit, thèse de l'Université de Savoie, 1991.

Mayère A., La gestion des savoirs face au nouveau modèle industriel, Revue Française de Gestion Ndeg. 105, sept-oct 1995.

McGuire J.G., Kuokka D.R., Weber J.C., Tenenbaum J.M., Gruber T.R., Olsen G.R., SHADE : Technology for Knowledge-Based Collaborative Engineering, Journal of Concurrent Engineering : Applications on Research (CERA) 1(2), september 1993.

Nonaka I., A dynamic theory of organizational knowledge creation, Organization Science, vol. 5, ndeg. 1, p. 14-37, february 1994.

Polanyi M., Knowing and being, The University of Chicago Press, 1969.

Pôle Productique, Opération Benchmarking Productique, Progresser en s'inspirant des meilleures pratiques, Pôle Infos, Pôle Productique Rhônes-Alpes, Ndeg.36, juin 1995.

Trivium, Commercial documentation of GINGO software, and press review on GINGO, Trivium, Paris, 1996.

Webber A.M., What's so new about the new economy ?, Harvard Business Review, january-february 1993.