Project : acacia
Section: Scientific Foundations
Keywords : Artificial Intelligence, Cognitive Sciences, Knowledge-Based System, Knowledge Acquisition, Knowledge Management, Knowledge Engineering, Knowledge Server, Corporate Memory, Ontology, Assistance to the User, Co-operation, Multiexpertise, Multiagent System, Conceptual Graph, Structured Document, XML, RDF, OWL, Semantic Web, Information Retrieval.
Knowledge Management (KM) is one of the key progress factors in organizations. It aims at capturing explicit and tacit knowledge of an organization, in order to facilitate its access, sharing out and reuse . The considered organization can be an actual enterprise or a public organization, but it may also just consist of a given department or service; it can also be a group, or a community, or a virtual enterprise (made of members possibly stemming from different companies, but sharing a common interest). An organization is made up of people interacting for common objectives, in an internal environment and with an external environment. These persons may have different functions and tasks in the organization, different competencies, knowledge, opinions, and work methods and they may produce explicit traces of their activities. In the course of their individual or collective tasks, they may need to find people able to give them useful information or to find such helpful information somewhere (in a document, a database, a CDROM, a film, etc.).
The members of the organization have individual knowledge (that may be explicit, implicit or tacit), as well as individual and collective objectives in the framework of their group or of the whole organization. The organization has global objectives and KM must be guided by a strategic vision. This vision enables to determine the main organizational objectives for KM:
Improve knowledge sharing and cooperative work between people inside the organization.
Disseminate the best practices in the company.
Preserve past knowledge of the company so as to reuse it.
Improve quality of projects and innovation.
Improve relationships with external world (such as customers, or privileged partners).
Anticipate evolution of the external environment (clients, competitors, etc.).
Be ready to react to unexpected events and to manage emergency and crisis situations.
So a KM policy must rely on a deep understanding of what is the organization, what is its corporate culture, what kind of knowledge exists (either individual, or collective in an internal group or collective in the whole organization), how can the organization's intellectual capital be assessed, how can the past explain the present and help to prepare the future, what can be the strategic objectives of KM and how they can be achieved according to the corporate culture and the environment of the end-users.
In an organization, knowledge can be individual or collective, it can be explicit, implicit, or tacit. In Nonaka's model , organizational learning relies on transformation between these different types of knowledge. Collective knowledge can also emerge in a community of practice. Tacit knowledge can be transmitted without any language (e.g. through observations), but in order to be transmitted to other persons, explicit knowledge generally needs a medium (i.e. document, database, etc.) so that people can create their own knowledge either by interacting with each other or by retrieving information from explicit traces and productions of other colleagues' knowledge. Knowledge can also be distributed among several knowledge sources in the organization, with possibly heterogeneous viewpoints.
There are three significant aspects to be tackled:
People (i.e. their knowledge, the way they acquire and communicate this knowledge, their organizational functions, their interest centers, their knowledge networks, their work environment, etc.): any KM solution must be compatible with the end-users' cognitive models and work environment.
Organization (i.e. its objectives, its business processes, the corporate culture, its corporate strategy, etc.): any KM solution must be compatible with the organizational strategy and culture.
Information technologies for supporting the intended knowledge management: the chosen technologies will depend on the KM objectives and on the intended end-users' environment.
The strategic vision for KM must enable to select the KM priority needs and to orientate the choice of relevant techniques. One possible approach for KM is the building of a corporate memory or organizational memory (OM). A corporate memory can be defined as an ``explicit, disembodied, persistent representation of crucial knowledge and information in an organization, in order to facilitate their access, sharing and reuse by members of the organization, for their individual or collective tasks''. So different scopes and grains are possible for an organizational memory. Its building can rely on the following steps (cf. figure 1) , with Management throughout all such steps:
Detection of needs in corporate memory,
Construction of the corporate memory,
Diffusion of the corporate memory,
Use of the corporate memory,
Evaluation of the corporate memory,
Maintenance and evolution of the corporate memory.
An organizational memory can be modeled from several perspectives: for whom, why, what, how, when, who and where. It aims at delivering the right knowledge to the right person at the right time in the right format, in order to enable the right action / decision. Although KM is an issue in human resource management and enterprise organization beyond any specific technological issues, there are important aspects that can be supported or even enabled by intelligent information systems. Especially artificial intelligence (AI) and related fields provide solutions for parts of the overall KM problem. Several techniques can be adopted for the building of an OM. The choice of a solution depends on the type of organization, its needs, its culture and must take into account people, organization and technology.
Several research topics can be useful for OM design:
Knowledge engineering and enterprise modeling techniques  can contribute to identification and analysis of a company's knowledge-intensive work processes (e.g. product design or strategic planning): the analysis of information flow and involved knowledge sources allows to identify shortcomings of business processes, and to specify requirements on potential IT support.
In order to acquire implicit knowledge, knowledge engineering methods and techniques are useful, in particular concepts handled in knowledge engineering such as ontologies, tasks and problem-solving methods. Knowledge modeling can be needed. The degree of depth of required knowledge modeling can vary: a significant depth can be required if the organizational memory is materialized in a knowledge base, a shallow modeling is sufficient for building a simple competence map of the organization.
Past experiments (e.g. lessons of past projects, past incidents, past successes or failures, etc.) can be represented in a case-based system ; case-based reasoning techniques can then be useful for retrieving them and reusing them for a new situation.
Ontologies can be a component of a corporate memory so as to be explored by the end-users; they can also be used for improving information retrieval about resources (such as documents or persons) constituting the memory if these resources are annotated w.r.t. the ontology. Such a use of ontology is close to the Semantic Web approach relying on metadata describing the semantic content of the Web resources, using ontologies . This approach for a corporate memory is inspired of the Semantic Web and is called ``corporate semantic web'' by the Acacia team , .
Natural language processing (NLP) tools can be exploited for the construction or enrichment of such ontologies  or for building annotations on the resources constituting the corporate memory.
KM in an organization requires abilities to manage disparate know-how and heterogeneous viewpoints, to make them accessible and suitable for the organization members that need them. When the organizational knowledge is distributed on several experts and documents in different locations, an Intranet inside the organization and Web technologies can be a privileged means for acquisition, modeling, management of this distributed knowledge. Agent technologies and Semantic Web technologies are a privileged way to handle such a distributed memory. Moreover, CSCW  offers an interesting way to enhance collaborative work between persons through distributed memories.
A specific kind of corporate memory is a project memory for preserving knowledge acquired during a project, for improving project management, for reusing past project experiences, design technical issues and lessons learned . KM can rely on the business processes. This process-oriented vision of KM can lead to OM integrating workflow systems.
A corporate memory can rely on a competence map, and techniques enabling expertise location are very useful for knowing who knows what in the company.
The Acacia approach relies on the analogy between the resources of a corporate memory and the resources of the Web. We consider that a corporate memory can be materialized in a corporate semantic web, that consists of :
resources (i.e. documents in XML, HTML or non Web-oriented formats, people, services, software, materials),
ontologies (describing the conceptual vocabulary shared by the different communities of the organization),
semantic annotations on these resources (i.e. on the document contents, on persons' skills, on the characteristics of the services/software/materials), these annotations using the conceptual vocabulary defined in ontologies.
The underlying research topics are:
How can we build and make evolve each component (resource, ontology, annotation)?
How can we build them semi-automatically through knowledge acquisition from textual sources or from structured database?
How can we take into account multiple viewpoints?
How can agent technology enable to build, manage and use a distributed memory?
How can we offer ``intelligent'', ontology-guided information retrieval or pro-active dissemination?
How can we rely on scenarios of use for needs detection and for stakeholder-centered evaluation?