The Nature of Enterprise Knowledge

August 26, 2009

In the 1990s most leading companies were very much concerned about knowledge management, recognizing that knowledge and competence are the driving forces of business, design and innovation. Industrial managers were also concerned about brain drain, loss of practical competence when skilled workers with multiple job experiences retired or left to join competitors.

This post is an attempt to revitalize industrial interest in KM by introducing new concepts and discoveries, such as knowledge architectures and families, and by giving good answers to industry questions like:

  1. What is enterprise knowledge?
  2. What inherent properties does enterprise knowledge exhibit, and what capabilities does it provide?
  3. How is enterprise knowledge best expressed, shared and managed by industrial users?
  4. How is work-centric knowledge best encoded to meet industrial needs?

The questions are answered based on scientific discoveries and experiences from industrial pilots.

Early Knowledge Management (KM) conferences attracted prominent speakers from many sciences, including Peter Drucker. Books were published on knowledge creation, processing and management, and research intensified on corporate memories. One may regard an Active Knowledge Architecture (AKA) as a visually operated corporate memory. Most conferences in those days discussed the core concepts, the basic principles, and definitions of what is knowledge and knowledge management. Many sciences contributed to the debate.

The surge for clarity, understanding, new approaches, methodologies and solutions culminated with the foundation of societies such as Knowledge Management Corporation International  (KMCI). Most societies set out to develop new knowledge management methodologies and knowledge management systems and frameworks. This has resulted in advanced information management systems utilizing the web, but so far no approach and solution that support design and innovation is industrialized.

Since the turn of the millennium the term Knowledge Management (KM) has practically vanished from the industrial scene. Most innovation projects are back to discussing improved Information Management using natural language semantics and search methods.

Forms and Kinds of Knowledge

The major forms of knowledge found in present enterprises are human mental models, document archives and computer application systems with databases. However, these forms are totally disconnected and their content is incomplete and uncoordinated. Only mental models are contributing to innovation and creative work, while document archives and databases rapidly become outdated and obsolete. They require manual management processes, which inhibit life-cycle knowledge support and reuse.

Enterprise knowledge is often viewed from business and engineering perspectives, defining knowledge areas and their interdependencies. Such an enterprise knowledge architecture is depicted in the figure below. It illustrates major kinds of enterprise knowledge, and the challenge of integrating the contents by natural language semantics and codes. The three layers could all contain conceptual as well as operational aspects.

In order to create coherent and consistent knowledge representations, enterprises must have access to services for holistic design, supported by integrated methodologies. Knowledge creation, sharing and reuse must become an integral part of practical work and should be performed in settings where context is easily captured and preserved. Knowledge expressed as static text, pictures, drawings, and models fail to capture contextual evolution and continuous change. They poorly support emergent work processes and situated learning. The meaning of terms, tasks, and even design rules change as creative and collaborative work is performed. This is why being able to execute emergent work processes in a modeling context is so important.

Kinds of Enterprise Knowledge

Kinds of Enterprise Knowledge

The dynamic modeling approach is in contrast to those who believe that ontology is an adequate base for model-driven systems engineering, and that local, situated and work-centric knowledge can be captured by IT systems. Ontology is best regarded as a reference model for design. It ought to be used by designers in order to maintain coherent community terminology and thesauri for enterprise information structures. Studies in epistemology and practical experiences have convinced us that work-centric knowledge should not be captured in natural language alone.

The Role of Mental Models

The KM World 1990 conference in San Diego was dedicated to discussing the questions: What is knowledge? What is knowledge management? Can we create and manage knowledge with current IT systems? The conclusions were that IT people have a lot to learn from pedagogy, epistemology, phenomenology and cognitive neuroscience. Design theory, systems engineering, and other practice-driven sciences should come together to develop standards, methodologies and languages. These approaches should allow designers to express and share fine-grained knowledge, supporting execution without programming.

Today associations like Building Smart in the construction industry have come a long way with standards and 3D rendering, but visual knowledge management for life-cycle collaboration and innovation has as yet not been given priority. 

The figure below was derived from discussions in international innovation projects in the early 1990s, involving leading pedagogues, design theorists and others with varying practical and scientific background.

Layers of Knowledge separated by Horizonz

Layers of Knowledge separated by Horizonz

The four horizons in the figure above are well recognized in pedagogy. They are typical of any enterprise knowledge, and relate to war-room thinking, design theory, practical work, phenomenology, and our knowledge spaces. These sources and manifestations of knowledge all require us to think and act in multiple dimensions. IT solutions should help us do so by augmenting our mental models with computer models and visual scenes of action. Proximity to action, in time, space and skill involved, determine the amount of work-centric knowledge created as tacit knowledge in our minds.

New visual approaches, integrated methodologies, extensible infrastructures, and user-managed services to express, share and manage work-centric knowledge will decide the future of industrial computing. Application systems off-the-shelf will never have user capabilities to support continuous innovation and accommodate local extensions and adaptations, and do not adequately support collaboration and community sharing.

Knowledge of what, how, who, where, when and why, for what and by what, and by and for whom etc. must be expressed as role-specific views in user interfaces adapted to support the tasks to be performed. If the views do not fit the roles and tasks, we will need view sharing, coordination and possibly create information ambiguity. Rich context to automatically support reuse can not be created purely by compiled code on top of a fixed data model.

Contributing Knowledge Areas

Major contributions to the evolution of active knowledge modeling technology have come from:

  • Product design for life-cycle support has been a major challenge for industry since the early days of Computer-Aided Design (CAD). CAD never supported conceptual or functional system design, and the recent extensions of CAD with KBE are just for automatic regeneration of design variants. However, design theory has provided the concept of a federated product model, replacing the many disjoint and typed product structures. Design theory has also contributed many modeling principles, such as making properties primary modeling elements.
  • Process engineering and work management have not yet been properly integrated. Process engineering has its roots in manufacturing of chemical and physical materials, energy production and other sectors. Process modeling comes top-down, while work management proceeds bottom-up. Now they are being linked into middle-out process and work management. This will enable concurrent resource allocation and management, task assignment and execution, work process monitoring and management, which is what is demanded by global industries. This is similar to how business processes should be modeled and implemented.
  • Systems engineering has provided evidence of the need for systems of systems, recognizing that we have systems for agreeing on the approach, roles and responsibilities, for integrating and adapting methodologies, for extending and adapting ICT infrastructures and finally for delivering product and services. This is true whether the product is hardware, software or brain-ware. The lesson to learn is that these systems should all represent reflective knowledge and be adapted in concert to provide complimentary services. 
  • Pedagogy and psychology of learning have provided most of our understanding of mental models and the functioning of our minds, and the recognition that design and innovation must involve situated learning. Cognitive learning as practiced in schools may be an effective way of learning for kids under 16, but for adults situated learning has many advantages to offer. 
  • Cognitive neuroscience is still a purely academic field, but under rapid development. Studying biological substrates and human processes, cognitive neuroscientists have provided a deeper understanding of how the human mind works in creating and using mental models. Neuroscientists have discovered that our long-term memory is managed by the motoric center of the brain. How our short-term memories feed the long-term memory has recently been discovered. This science would benefit from visual modeling and active knowledge architectures. 
  • Epistemology taught us that the meaning of terms change with time and context. Ontology, thesauri and other structures consisting of terms are important, but primarily for communication with outsiders. Knowledge categories and families reduce the need for precise terminology by allowing context- and role-specific classifications.
  • Sociology provided knowledge required to understand networking, team-building and collaboration across borders. The principles of role-orientation with clear division of responsibilities, formation of effective work environments, and needs for tailoring services have been inspired. 
  • Natural language has its powers in aligning subjective human perspectives to create objective models, in social learning. However, contextual, work-centric knowledge is only shared through joint participation and experience, not by writing and talking.  
  • Enterprise modeling languages provide richer, participative ways of capturing context, of externalizing and sharing tacit knowledge, and of capturing design intent. Natural and graphical languages have complimentary properties and capabilities. 
  • Phenomenology has supported the understanding of knowledge spaces and new multi-dimensional approaches to enterprise modeling, introducing new fine-grained modeling languages.  Modeling methodologies must become operational and provide inline user guidelines to widen our perception and understanding of collaboration. 

Work-Centric Knowledge

The concept of work-centric knowledge is even more valuable than situated knowledge as it is role- and task-specific, and can be directly fed into an execution engine. Some definitions of knowledge are quoted below. The colors relate to the knowledge layers in the previous figure. Work centric knowledge, a flux of views among collaborating humans, is the only definition that includes externalization and activation of tacit expert knowledge.

 

Definitions of Knowledge

Definitions of Knowledge

 Capturing and Sharing Work-Centric Knowledge

Enterprise knowledge processing, expressing, sharing and managing knowledge for execution without programming, has lead us to characterize work-centric knowledge as knowledge with continuously improving properties and capabilities. Collaboration and management of creative work require active, visual models to close the gaps between design and execution. This means that models need to be updated and extended as project work progresses.  The figure below illustrates the transition from data to new methods that knowledge architectures support. The transitions, indicated by the peripheral arrows, may be performed by adding reflective views with relevant content, avoiding massive information management.

The Work-centric Knowledge Management cycle

The Work-centric Knowledge Management cycle

The most valuable knowledge to any project is work-centric knowledge, simply because it represents your core competitive advantage and innovative power. However, as projects involve customers, suppliers and others there will be a need for capturing work-centric knowledge from many players. However, the work-centric knowledge of a supplier may be best presented as participative or even sense-making knowledge to other project participants. The layers of knowledge, separated by horizons, are illustrated in the figure below (illustrating the depth below the surface of the previous figure).

Core Enterprise Knowledge

Core Enterprise Knowledge

Active knowledge architecture is a way of interrelating, cultivating and managing enterprise knowledge, enabling roles to define appropriate views and associate knowledge contents needed to execute tasks.

Enterprise Knowledge Spaces

War-room thinking surfaced in US industry in the 1980s, and was brought to Europe by the automotive industry some years later. In the late 80s projects were launched where the goals were to develop dynamic visual models of products. These projects developed the first prototypes of enterprise knowledge architectures, and combined with the lessons learned from pedagogy, sociology and phenomenology a solid foundation of concepts and principles for the various enterprise knowledge spaces were developed.

Over the last decades, European research projects have made many important discoveries and implemented many innovative designs. Aspects of active knowledge architectures can be traced back as far as 1994. The most important discoveries belong to these four knowledge areas:

  • Visual collaborative scenes enabling proactive behavior
  • Closing the gap between product design and IT implementation
  • Complementary roles of mental and computer models
  • Classes, categories and families – definitions and values

According to phenomenology of perception, knowledge space boundaries may be caused by the limitations of current media (information carriers), modeling languages and possibly our mental processes.

Capabilities of Work-Centric Knowledge

Work-centric knowledge has four intrinsic capabilities: Reflection, recursion, repetition and replication, also referred to as 4R.

Reflection is illustrated in the figure below. A good example of reflection is the interplay between an athlete and his trainer. The trainer, responsible for the how process, needs to see the athlete perform the what process in order to communicate an improved performance, and the athlete needs to see the trainer demonstrate the improved performing approach.

 

Reflective Work-centric Knowledge

Reflective Work-centric Knowledge

 Recursion is used in product design and the processes of balancing parameter values among engineering disciplines responsible for different systems. If a car designer changes the front fender, the coefficient of wind resistance will change. This may cause changes to all car systems. Recursion enables automatic recalculation of the system parameters impacted. Repetition is also known from industrial practice, being used to save successful industrial processes, and repeat them in later projects. Replication is used to spread content or copy and share content or whole structures. Successful quality content is most easily replicated by use of reference models and templates.

Knowledge Management by Work Execution

As late as 2002, at the end of the European 5th Framework Program, enterprise modeling experts were still discussing the development of enterprise typologies, built as type hierarchies, as the way forward for handling enterprise model and language elements. This would imply that all types must be created and managed in levels of public and private hierarchies. Modeling languages would use objects, relationships and predefined  artifacts from these type hierarchies.

The importance of the granularity of modeling elements, discovered by the CIMOSA project is the key to be able to execute models. The Information, Roles, Tasks and Views (IRTV) language is used to model the executable aspects of products, organizations, processes and systems. Relationships in early design are modeled as tasks to facilitate re-iteration and recursion. This is illustrated by the arrows in the figure. By this approach the amount of modeling is drastically reduced, because there is no need to replicate execution details in every product, organization, process, or system model.

Enterprise Languages, Modeling and Execution

Enterprise Languages, Modeling and Execution

Modeling in the IRTV language has allowed us to close the gaps between design and execution. It also means that knowledge cultivation, design language composition and holistic approaches can be implemented by practitioners. Projects can design and manage their own purposeful languages and architectures and create adaptive product, organization, process and system models.

Classes, Categories and Families

Classes are sets of objects that have common internal properties. Industry, long before object-orientation was introduced to IT, had developed their own object classes, e.g. industrial part catalogues.

Categories consist of objects that share the same functional properties, and perform the same functions. Categories are useful for managing modules and functional systems. They integrate components designed by separate suppliers. Categories are defined by a combination of methods and rules, which can be flexibly implemented as task-patterns and module-specific parameter-trees.

Families are groups of objects that satisfy common requirements, functions, features, performance parameters, and design principles. With families, solutions can be configured by applying design and configuration rules, automatically selecting contents from relevant categories and classes. Knowledge families will allow semi-automatic life-cycle knowledge management and reconfiguration of solutions, and thus ease the planning and execution of repairs, modifications and updates.

Classes for global reference and replacement are independent of roles, while categories are often role-dependent. Effective use of families is dependent on customer and client roles and services for automatic configuration of replacements and modifications..

The combined capabilities of knowledge classes, categories and families and their effect on industrial computing and knowledge management are still a matter of research and industrial piloting. The notion of automatic customized design and adaptive life-span architectures, systems and solutions could soon be reality.

Complimentary Mental and Computer Models

Back in the 1980s many people were discussing the rational division of work between humans and computers. Knowledge management, artificial intelligence, smart products and industrial war-rooms were discussed at length. Questions included: What is best managed as mental models? What goes into externalized models managed by computers? In the following table we compare the strengths, capabilities and properties of mental and computer models.

Mental Models Computer Models

Best performance processes

Conceptual creativity and ideation Engineering calculations and computations
Imagination, fantasies and dreams Standardization, modeling and simulation
Reasoning, separation and discrimination Dynamic viewing and presentation
Socializing, arbitration and consensus making Communication, collaboration and coordination
Association and recollection Multidimensional rendering
Comparison and selection Configuration management of complex structures
Holistic consideration and focusing Handling families, categories and classes
Validation, usability and concern resolution Handling complexity, rules and dependencies
Prioritization and decision-making Storage and retrieval
Handling informal pragmatic logic Handling formal, scientific logic

Viewing capabilities

Subjective perspectives Objective and systemic views
Combining perspectives and operational views Interactive presentations
Selective focusing and filtering Dynamic collection and sharing
Feature and action-driven Sustainable composition and contents

Properties

Ambitious and talented, uncertain Consistent and reliable
Impulsive and innovative Capacities and speed of processing
Reflective and recursive Persistent, repeating and replicating
Proactive Reactive

An active knowledge architecture can combine the best from both worlds, and computer models may augment our mental models. Combining them could also have serious impact on cognitive neuroscience and phenomenology.

Conclusions

Now the four initial questions should be answered:

  1. What is enterprise knowledge?       
    The answer depends on proximity to action, the scene of action, and experiences from similar settings. The human mind is a very powerful conceptual processor. This capability is not yet supported by computer models.
  2. What inherent properties does enterprise knowledge exhibit, and what capabilities does it provide?    
    We have briefly talked about reflection, recursion, repetition and replication, and the capabilities it provides for continuous innovation and for automating enterprise knowledge management.
  3. How is enterprise knowledge best expressed, shared and managed by industrial users?       
    The best expression of work-centric knowledge is as a scene of action where the 4Rs are supported, enabling designers and engineers to continuously express local details and execute methods.
  4. How is work-centric knowledge best encoded to meet industrial needs?       
    Our pilot implementations have convinced us of the powers of the IRTV language, and of the ability of role-specific model generated workspaces to capture context and automate knowledge management.

Industrial Computing in 2012 and 2020

Method engineering and role-orientation, exploiting classes, categories and families, all implemented in active knowledge architectures, may set the future direction of industrial computing

Active knowledge architectures close the gap between model design and model execution. Pilots of visual scenes for collaboration have already been successfully implemented. This indicates a paradigm-shift in computing. This shift will benefit human creativity and collaboration, and put more emphasis on model-driven solutions, local practices, learning communities, and on the job training.

By 2012 we should have the first industrial solutions covering the entire project life-cycle rolled out, and experiences can fully validate the Active Knowledge Architecture approach.

By 2020 industry will no longer buy off-the-shelf application systems. Continuously adapting system landscapes will be fully operational, supported by model-driven computer environments, enabling industry to learn from operations and practical work environments.

The value of active knowledge architectures, supported by object classes, method categories and solution families are not yet widely understood, accepted, implemented and validated. Only industrial experts willing to share their knowledge, competence and local skills can realize the powers of active knowledge architectures. It cannot be bought off-the-shelf or out-of-the-box from any vendor.

  

Advertisements

2 Responses to “The Nature of Enterprise Knowledge”


  1. […] The Nature of Enterprise Knowledge « Active Knowledge Modeling (tags: km) […]


  2. […] work-centric and situated enterprise knowledge demands fundamental rethinking of not only the nature of enterprise knowledge, but also the practical approach to realizing knowledge management. Knowledge workers must be […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: