One of the most serious impediments to cost-effective use of knowledge is that often knowledge bases or systems are constructed afresh. It is unusual for problem-solving experience or domain content to be acquired and then reused, partly because knowledge tends to require different representations depending on the problem-solving that it is intended to do. Understanding the use and application of knowledge would enable more leverage to be gained from the knowledge already at hand, thereby increasing the returns on the investment in those knowledge assets.
In knowledge reuse, the aim will be to exploit the software engineering principle of reuse as much as possible. An obvious place to start will be the field of knowledge engineering, in which there are basically three reusable types of object. First, there are ontologies. Second, there are problem-solving methods (PSMs), which describe ways of solving particular problems, and specify the types of knowledge that will be required for the method to work. PSMs are generic objects - hence reusable - and when knowledge of the requisite type is acquired and used by the method, the result is a domain-specific description of the problem-solving in the domain. Thirdly, there is also the possibility of reusing knowledge bases (KBs) themselves, and we shall be looking at ways of adapting existing KBs for use in other contexts.
Of course, not all reusability issues are based around knowledge engineering. For example, it is possible to identify user profiles, such as navigation strategies for getting round a complex website. Reuse of such strategies is very likely to be a valuable extension of our web technologies.
In many cases, reusing knowledge is non-trivial. This counterintuitive result is generally due to two particular problems. First, in an organisation of more than moderate complexity, locating the knowledge to be reused is difficult. Distributed agents may be unaware that the knowledge they need is available; this is the challenge of knowledge retrieval. Secondly, the knowledge may be held in the organisation and correctly identified, but may simply be in the wrong form for the task - the essential information may be only implicit in the repository. The knowledge may have to be reconfigured in some way to meet the requirements of the task in hand. It may be that the knowledge requires some partial modification (e.g. updating). Here, understanding the knowledge requirements of the task is the key to understanding, identifying and using the correct knowledge from the various sources.
AKT Technologies addressing issues in Knowledge Reuse --