When a knowledge repository gets very large, finding a particular piece of knowledge can become very difficult. There are two related problems to do with knowledge retrieval. First, there is the issue of finding knowledge again once it has been stored, understanding the structure of your archive in order to navigate through it efficiently. And second, there is the problem of retrieving the subset of content from the repository that is relevant to a particular problem. This second problem, the dynamic extraction of knowledge from a repository, may well set problems for a knowledge retrieval system that alter regularly and quickly during problem-solving.

In knowledge retrieval, the aim would be to develop user-friendly tools for retrieving knowledge from repositories. One obvious place to begin is to try to exploit natural language. Natural language can be used as the basis for the interface to our knowledge services. For instance, ontologies could be used to interpret knowledge queries in natural language forms from users.

The other focus of our effort to meet this challenge is focused around the technologies behind the interface. Our expertise in a number of areas, such as search engines, multimedia thesauri, capability decisions and statistical sampling will be used to increase the efficiency of retrieval technologies, and to provide technologies that cope with the new demands of the semantic web.

In particular, we are very alive to the technological possibilities of the enriched markup of web pages. Given the central importance of the Internet, web pages will clearly be a vital storage and dissemination medium for knowledge. The use of semantic markup technologies such as XML or RDF, driven by our proven knowledge technologies such as ontologies and dynamic hypertext link generation, will enable efficient and accurate retrieval of web pages from the web itself, or from highly structured and possibly very complex intranets.

When knowledge is stored or retrieved, it is important to remove any duplication to avoid overloading users with knowledge. AKT is developing a variety of knowledge fusion techniques to identify and remove such duplications before presenting knowledge to the user.

AKT Publications addressing issues in Knowledge Retrieval --