Although we often suffer from an overdose of data, all too often the problem is that the knowledge available is insufficient or poorly-specified. The challenge here is to get hold of the information that is around, and turn it into knowledge by making it usable. This might involve, for instance, making tacit knowledge explicit, identifying gaps in the knowledge already held, acquiring and integrating knowledge from multiple sources (e.g. different experts, or distributed sources on the WWW), acquiring knowledge from unstructured media (e.g. natural language or diagrams).

Knowledge acquisition (KA) is a field which has reached a certain level of maturity. It began as part of the drive to build knowledge-based systems, and was a line of research devoted to developing methods and software tools to provide knowledge content for such systems. There are many tools and techniques available, and a number of integrated workbenches and methodologies on the market. It is not the intention of AKT to reinvent this particular wheel. However, where AKT can make a difference is in a number of tricky, non-mainstream areas of knowledge representation from which the extraction or acquisition of knowledge is non-trivial or poorly-understood.

Ontologies, or sharable conceptualisations of domains, are a central technology for knowledge management (KM). However, ontology construction is a considerable overhead on any KM programme. AKT is investigating generating ontologies automatically from text, using text mining and natural language techniques. Such technology is expected to be very important given that much knowledge in science and industry is kept in informal natural language repositories. AKT is expected to make progress in the integration of natural language information extraction (IE) techniques with standard KA methods, for example applying domain ontologies to facilitate the IE from texts.

Other areas of interest to AKT are: artefact enrichment, in other words, methods, languages and logics for annotating and enriching the content of objects such as web pages, KA materials, hypertext fragments etc.; multimedia KA, i.e. coming to understand (finding models and structures for) knowledge expressed in non-text media; and incidental KA, the acquisition of knowledge as a by-product of other processes. For instance, software to facilitate meetings and collaboration can also help with integrating multiple perspectives and the construction of a collective memory resource. Finally, the field of KA can be transformed from its current state by a shift in the technologies available for it. There are two obvious directions of research to exploit here. First, there is the harnessing of the Internet. As noted above, there have been integrated sets of KA tools in workbenches, but such integration can be carried a step further by integrating the tools in a web-based environment, thereby allowing, e.g. the use of web-based libraries of knowledge model components, or the generation of knowledge models in XML.

The second direction of technological change for KA is that of software agents. Much KA, e.g. from information repositories on the WWW, might be automated by using intelligent agents, armed with structuring schemata such as ontologies of the domain in question.

AKT Publications addressing issues in Knowledge Acquisition --