Combining data mining and human expertise for making decisions, sense and policies

Abstract

A major challenge facing management in developed countries is improving the performance of knowledge and service workers, i.e. the decision and policy makers. In a developing country such as South Africa, with a well‐developed business sector, this need, especially in government, is even more crucial. South Africa has to face many new challenges in the 21st century ‐ growing environmental concerns, massive social and economic inequalities, high occurrences of HIV, low productivity, massive unemployment and the nation’s evolving role in Africa, amongst others. The importance of a sound science and technology policy framework to address these pressing issues cannot be overemphasised This paper discusses the construction of a knowledge‐base from a data repository concerning a South African National Research and Technology (NRT) Audit. This knowledge‐base is to be used as an aid when developing a science and technology policy framework for South Africa. The knowledge‐base is constructed using the cooperative inductive learning team (CILT) approach, which combines diverse data mining tools and human expertise into a cooperative learning system. In this approach, each data mining tool constructs a model of the knowledge as contained in the data repository, thus providing an automated tool to make sense of the knowledge embedded therein. That is, the data mining tools learn from the data in order to obtain new insights. The system also incorporates human domain expertise through the computational modelling of the human subject knowledge. The knowledge, as obtained during team learning, is stored in a team knowledge‐base. Results indicate that the CILT learning team approach can be successfully used to make sense of the vast amounts of data collected and provide a knowledge repository for further decision making and policy formulation.