What do we know and how can we manage our knowledge assets to fulfill our organizational objectives? These are important questions for an organization. Knowledge management should enhance individual, group, and organizational learning; improve information circulation; and even support innovation. It aims to capture and represent an organization's knowledge assets to facilitate knowledge access, sharing, and reuse. In doing so, it attempts to address these possible goals: capitalizing on individual know-how in a collective knowledge; improving newcomer learning and integration; disseminating best practices; improving corporate work processes, product quality, and productivity; and reducing new product design times. We must tackle these complex problems from at least the human, socio-organizational, and technical viewpoints.
A corporate or organizational memory is 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. 1,2
That organization can be an actual enterprise, a public organization, or just a department or service. It can also be a group, community, or virtual enterprise comprising members from different companies, gathered by a common interest.
Different scopes and grains are possible for a corporate memory, but its development relies on these steps: 2
detection of needs, construction, diffusion, use, evaluation, and maintenance and evolution. We can model the organizational memory from several perspectives; for example, John Kingston and Ann Macintosh emphasize the what, how, when, who, where, and why perspectives. 3
Fundamentally, an organizational memory aims to deliver the right knowledge to the right person at the right time in the right format to enable the right action.
Techniques for building an organizational memory can be noncomputational, database-, document-, knowledge-, case-, or Web-based; 2
they can also be product- (based on repositories) or process-oriented (based on corporate work processes). The choice depends on the type of organization, its needs, and its culture and must take into account people, organizations, and technology. (See the sidebar for a description of some of these techniques and articles in this special issue
that use them.)
Exploiting the Internet and Intranets
Individuals and organizations can take advantage of the remarkable possibilities of access to information and knowledge that the Internet provides. Web technologies such as HTTP and HTML have dramatically changed enterprise information management. Moreover, an intranet relying on Internet technology and protocols enables intraorganizational communication and internal information sharing through the corporate internal network. For example, a multinational corporation can benefit from intranets and the Internet to gather, manage, distribute, and share knowledge, inside and outside the corporation. Figure 1
shows the role of the Internet and intranets in the corporate memory management cycle.
Figure 1. Exploitation of the Internet and intranets in the corporate memory management cycle. 2
Corporate Memory Creation and Revision
A company can exploit the Internet and intranet features in several ways. It can use internal HTML or XML pages or external URLs containing organizational memory, making it accessible throughout the company. Internal or external discussion forums, newsgroups, news archives, and FAQs also encourage information exchange.
More proactive methods of creating and revising corporate memory include integrating messages exchanged through e-mail in the corporate memory, extracting information from the external Web sources for technological or strategic intelligence, and using computer-supported cooperative work (CSCW) tools to support complex-system collaborative design or collaborative software development.
The wide variety of organizational choices involves several actors with different roles:
• human knowledge sources (such as experts, specialists, or operators), whose knowledge must be made explicit or who have written documents that others will access through the organizational memory;
• knowledge engineers, who acquire and model knowledge;
• knowledge watchers, who gather, filter, analyze, and distribute knowledge elements from the external world (from external information Web sources, for example);
• organizational memory developers, who concretely build, organize, annotate, maintain, and evolve the corporate memory;
• a team of validating experts (for example, a reference team), who validate the knowledge elements before their insertion in the organizational memory;
• corporate memory users, who must easily access and reuse memory elements; and
• organizational memory managers, who supervise the organizational memory project.
External information sources
A corporate memory can rely on the internal competencies inside the company and on external information sources that provide knowledge from the external world—that is, useful for enterprise activities. Knowledge watchers collect scientific, technical, and economic information from journals, newspapers, and the Web. They then filter, analyze, and validate that information to integrate some interesting elements into the corporate memory or forward them to the relevant members of the company. An intranet can exploit an internal corporate memory, while an external memory can rely on either an extranet connecting the company and some privileged partners such as customers, suppliers, and subcontractors or the Internet and the Web.
The different kinds of intelligence interesting to a company include
• technological intelligence, to follow an existing or an emerging technology;
• competitive intelligence, to know about activities, products, or services of competitors or other actors in the enterprise market;
• commercial intelligence, to know about the enterprise commercial environment such as distributors, suppliers, and customers; and
• strategic intelligence, to support the enterprise managers' strategic decisions.
Agents monitoring news or Web information can gather and filter information for technology watching. Companies can also exploit research on intelligent integration of information from possibly heterogeneous, voluminous sources of information on the Web.
Corporate memory organization
In the framework of an organizational memory materialized in documents (or informal knowledge), we can associate such documents with a formal knowledge on which we can perform a reasoning to retrieve adequate parts of the document. This formal knowledge can either represent a part of the document or consist of metaknowledge about the document—possibly with more knowledge than the document includes. (Several articles in this special issue
use this approach; see the sidebar for examples.)
For example, Ontobroker exploits an ontology to guide information retrieval from annotated HTML documents accessible on the Web. 4
This approach aims to improve classic Web search engines with semantics-based information search capabilities and relies on semantic metadata or annotations on the resources constituting the corporate memory. Some research on text-mining techniques aims to automate partly the building of such semantic metadata, using textual-document automatic analysis based either on statistical or linguistic techniques. Recent research on the semantic Web 5
can have a significant impact on corporate knowledge management.
Dealing with documents in multiple formats is often needed because not all companies impose a uniform document format. In this instance, storing short document descriptions (or semantic metadata about them) with hypertext links in the organizational memory instead of the whole documents is a possible solution. 6
Another solution is to wrap company documents in XML documents. Moreover, ongoing research on the Resource Description Framework could let us express semantic metadata on the resources constituting the corporate memory.
In the future, thanks to XLink, we will be able to exploit sophisticated hypertext links. We can implement different types of presentation of the same document, which lets a company present an XML-based corporate memory differently according to user profiles. For example, the disciplines, professions, or roles of the project participants or the interests and contexts of the potential users can determine organization memory views. However, organizing the corporate memory through static views related to the knowledge sources and potential user profiles might be insufficient. A knowledge management system can also dynamically create the corporate memory view when a user tries to access the organizational memory in a given context, provided that the system can access a user's profile or a task description.
Cooperatively creating and revising
Several members of a company can use the Internet to cooperatively create and reuse organizational memory. In this case, distributed editions of the corporate memory must be possible and the consistency and security problems arising from this distant edition must be solved.
According to the organizational choices, the corporate memory creation and evolution can be distributed (for example, each knowledge source can directly update the memory) or centralized (for example, by a person or a service that manages the corporate memory). A distributed corporate memory can support collaboration and knowledge sharing between multiple people in an organization or in several collaborative organizations, even if they are geographically dispersed. It is particularly useful for dynamically building and preserving the memory of an ongoing project. The explicit representation and visualization of argumentations exchanged during team discussions let the team members collaboratively capture an ongoing project memory.
A particular case of project memory is a design project memory. Companies can use these techniques to keep a memory of software development projects and of complex system design projects and might let designers exchange knowledge both on the design products and on the design processes.
Because a corporate ontology might be part of an organizational memory, 7
companies can exploit tools for cooperatively building a consensual ontology. 8
Examples of such tools are the Ontolingua server, 9
and WebOnto. 11
Ontologies can help index the organizational memory and to enable later semantic searches and knowledge retrieval in the corporate memory materialized in documents or in news archives. They are also useful for pushing relevant information to the relevant users. When the organizational memory is materialized in a knowledge base, tools supporting knowledge acquisition, modeling, and inference through the Web, such as WebGrid, become important. 12
Knowledge diffusion lets enterprises take advantage of popular Internet services such as e-mail and the Web. The Web can be a basis for uniform information distribution, independent of how you store information. Diffusion can thus rely on populating knowledge elements on the Web or on a knowledge server on the Web. 8,13,14
Users can access documents (such as HTML and XML documents), relational or object-oriented databases, ontologies, knowledge bases, case bases, digital journal articles, and news archives over the Internet. Therefore, companies should consider multiple knowledge servers: document, ontology, knowledge-base, database, electronic journal or digital library, or news.
The main problems are
• organizing and possibly indexing the corporate memory to enhance its diffusion;
• retrieving relevant elements of the corporate memory to answer a user's request or proactively push relevant elements toward users; and
• adapting the answer to users, in particular to their tasks, according to the corporate work processes.
Such problems seem similar to those of information retrieval from the Web, but they are specific to an organizational-memory framework. Therefore, solutions such as Web search engines or Web intelligent agents might be useful but must be adapted in the context of a corporate memory. For example, they can rely on a domain ontology or an enterprise model to improve information search. In addition to the numerous innovative applications of the Web, 15
we must take into account human factors because using the Web involves new types of human-computer interaction and human-human cooperation. 16
In some companies, an intranet can be the tool for a corporate reorganization, aiming at a less vertical and more horizontal organization, because it can enable information distribution at several (or even all) hierarchical levels.
One goal of knowledge distribution is to improve individual, group, and organizational learning. Several kinds of members of a company might be interested in knowledge distribution, such as managers, designers, engineers, and employees.
Knowledge might also be disseminated to external people (clients, providers, and privileged partners). Virtual enterprises can distribute knowledge to different organizations with different corporate cultures, which might be important in multinational corporations.
Knowledge distribution can be active (for example, using push technologies such as proactive agents) or passive (using pull technologies). 1
Companies must adapt the distribution to their users' daily work environment.
This special issue illustrates current research in knowledge management. I hope it will stimulate researchers to deepen the links between knowledge management and the Internet and intranets. The world-wide, interdisciplinary research on the semantic Web 5
can find a killer application in knowledge management. Knowledge management will benefit considerably from research, technological progress, and innovative Web and intranet applications, provided that such technology is integrated in the organization to favor knowledge sharing.
I thank the reviewers for their careful reviews of the 35 articles submitted to this special issue. I also thank the Acacia team members, the Memento working group participants, and the IST Comma project members for their fruitful discussions on knowledge management, the Internet, and intranets.
is research director at Inria Sophia Antipolis and scientific manager of the Acacia Project at Inria Sophia Antipolis. Her research interests include knowledge acquisition, knowledge modeling and management, conceptual graphs, and multiagent systems. She has published more than 50 articles in journals, books, or conferences and organized 15 conferences or workshops. She has a PhD in computer science from the University of Paris-Sud, Orsay. Contact her at Inria, Acacia Project, 2004 Route des Lucioles, BP 93, 06902 Sophia-Antipolis Cedex, France; Rose.Dieng@sophia.inria.fr.