Chapter II: Knowledge Management
In this chapter, we will explore the field of KM in order to set the context for the study. We will examine the tools and techniques for the management of knowledge from a historical perspective and also explore the issues for managing knowledge in a spatially distributed environment. We will see that there is a tension between different views of knowledge and, for the purposes of this book, we will regard knowledge as "soft" and "hard" as a working definition.
We have already seen in Chapter I that commercial organisations have come to recognise the value of the knowledge that is held in the organisation. As knowledge has come to be recognised as an important resource and a valuable asset, it has also been recognised that it must be managed as such. KM is a rapidly growing field that is dismissed by some as simply the most recent management trend but recognised by others as essential if organisations are to cope with pressures brought about by downsizing, outsourcing, and globalisation.
There is a wealth of definitions of KM available, most of which will have different meanings for different people. However, if something is to be managed, many people feel it must be able to be quantified, counted, organised, and measured (Glazer, 1998), and it must be able to be built, owned, controlled, and its value maximised (Allee, 1997). This view of "management" has influenced attempts to manage knowledge and led to attempts to quantify, capture, and control it as an object. More recent developments in KM have demonstrated that this approach to the management of knowledge is too restricted and that some aspects of knowledge cannot be captured. Therefore, the concept of KM has had to be broadened to include elements of sharing, learning, the generation of new knowledge, and the application of knowledge.
MANAGING KNOWLEDGE FROM THE HISTORICAL PERSPECTIVE
In Chapter I we saw that organisations have lost personnel through downsizing and outsourcing activities. As staff have left the organisation, they have taken knowledge with them. As a result of this, initial attempts at KM consisted of trying to "capture" knowledge in order to be able to use it when people had left or at least to be able to pass it to new people who may be less skilled than their predecessors.
This approach views "knowledge as information" and as an object that can be codified. This led to attempts in the 1980s using AI to extract the knowledge of experts in order to create "expert systems." Von Krogh (1998) refers to the aim of this approach as being "to create information-processing machines that would resemble human intelligence. These machines would, like the brain, manipulate symbols and thereby solve predefined problems" (pp. 148-149).
The difficulty of this approach was demonstrated by Roschelle (1996), who explored the use of computers in learning. He was particularly interested in whether representing the mental models of experts was sufficient to help novices. As a result of his studies he came to the conclusions that:
Gaps between world views prevent students from interpreting displays literally, and thus limit the extent to which communication can be achieved by representing knowledge accurately. Hence, rather than merely representing mental models accurately, designers must focus on supporting communicative practices (p. 15).
The expert systems failed to live up to expectations and only a few of the expert systems developed in the 1980s were still in use in the early 1990s (Davenport &Prusak, 1998). Part of the reason for this may well be the exaggerated predictions made for such systems, which perhaps led to unrealistic expectations. The expert systems attempted to capture knowledge as explicit decision rules. Most of the systems did not meet expectations because decision rules have a limited scope and are not able to cater to rapidly changing environments (Davenport & Klahr, 1998), and because knowledge is not like this.
Berg (1997) describes the difficulties in developing and implementing two medical expert systems and observes that the systems did not work as intended. That is, in order to function at all, the practice in the environment had to be changed, and the systems failed to fulfill the original expectations of the developers in terms of power and scope and also the original aim of singlemoment intervention. The systems had to become localised—both in scope and in space. They were localised in scope in that the original hopes for the system had to be drastically scaled down, and they were localised in space in that they did not transplant to other environments, both in terms of a more universal application and in terms of the same application in a different environment.
Despite the failure of expert systems to live up to expectations, the view of knowledge as an object that can be captured continues to dominate the KM field, with some researchers still viewing the capture of knowledge as the main challenge (Alavi & Leidner, 1997). Work continues in AI under the heading of Knowledge Engineering and using ontologies (Buckingham Shum, 1998; Vasconcelos, 1999), and Case-Based Reasoning (CBR) (Davenport & Klahr, 1998).
Later attempts at KM continued to view knowledge as an object, but the emphasis shifted from trying to encode knowledge into expert systems to the creation of systems for capturing it and storing it for subsequent sharing. This is still a view that is prevalent in many definitions. This approach continues to focus on the capture-codify-store cycle and attempts to separate knowledge from people. This approach views knowledge as something which can be abstracted and which is embodied in documents (for example, books, reports, presentations) and which can be stored using databases, knowledge bases and data warehousing.
There are, however, successful examples of this approach to KM. In attempting to manage knowledge in line with the view of management as quantification and control, many companies looked for tangible examples of what they termed their "intellectual assets," such as patents, copyrights, trademarks, and documents. Allee (1997) and Cohen (1998) both cite the example of Dow Chemical. Dow Chemical concentrated on screening patents to ascertain which should be exploited, licensed, or abandoned. This area was previously under-explored and exploiting it has resulted in $40 million savings over ten years and $125 million in licensing income. Other examples are of companies creating organisational directories or "Yellow Pages" of staff members with specific expertise. Others speak of the organisational knowledge in manuals, databases, document repositories, and filing cabinets. This is the most common use of technology in KM—to create a repository of structured or explicit knowledge (Davenport & Prusak, 1998), but this has more in common with Information Resource Management (IRM). Indeed, IRM is a necessary part of KM, but, in some cases, what is presented as being KM is simply IRM with a new label. This is reflected in the offerings of Knowledge Management Systems (KMS). Many of these are marketed as being total solutions for KM needs, but in fact many are simply the vendor's information retrieval system that has been given a new name to capitalise on the KM market.
Unfortunately the total KMS does not exist. As KM is not a single simple process but a range of processes, the needs of KM are too broad for any one system to be able to cater to them all. More important, however, is the recognition that IT is not a solution but an enabler and a support.
Despite this recognition, there is still a bias in the trade literature to approaches that view knowledge as an object. As a result, many organisations still equate knowledge with information and regard it as an object that exists on its own and can therefore be captured, stored, and transmitted. This leads to a technology driven approach to KM, as discussed by Fahey and Prusak (1998):
This orientation is in turn reflected in and reinforced by the pervasive information technology approach to the management of data and information: capture, store, retrieve, and transmit. Although organizations obviously need to manage their data and information using these technology-centered models, knowledge is a substantially different thing and thus needs different models (p. 267).
From all this we can see that in the management of knowledge from a historical perspective one specific view of knowledge has been prevalent, that is, as a commodity that can be represented, stored, and manipulated. As a result, two primary routes have been followed:
To capture the knowledge structure or mental model of an expert and encode it into an expert system. This was an approach that did not live up to expectations.
To capture, store, and disseminate knowledge, for example, in books, databases, and reports. This has led to a need to cope with information overload and has more in common with IRM than KM; however, it does demonstrate that IRM is an essential part of KM.
CBR is a form of AI but instead of using ‘classical’ rules, the system's expertise is held (or embodied) in a library of past cases. Each case holds a description of the problem. It will also record the solution or the outcome. If a user has a problem (s)he matches their problem against the cases in the case base and retrieves similar cases. These are then used to suggest a solution. The solution is tried and revised where necessary. This new case will then be entered as a new case to the case base.
Some people consider a knowledge base to be part of an expert system and that it contains facts and rules which are needed for problem solving, however, the term is now more widely used to mean that it is a searchable electronic resource (which may simply be a database) which is used for the dissemination of information, often via the Internet or an intranet. It may be accessible to the public. An example is the Microsoft knowledge base which is available on the Microsoft support site. As problems arise and are solved they are written up and entered into the knowledge base. Any Microsoft user can access the site and type in their problem query. The system should then return a number of possible solutions—all without the need to contact a technician.
A data warehouse is a consolidated view of enterprise data. The data will have come from a variety of sources (production and transaction data). The data warehouse will therefore contain a wide variety of data. The data are optimised for reporting and analysis and are designed to support management decision making. As the data are from a variety of sources but are optimised they present a coherent picture of business conditions at a single point in time. It is easier to run queries over data which have been warehoused than over data which are in several different locations and systems.