Make: Maryland Automatic Knowledge Extractor

This paper discusses the need for developing an automated process for knowledge acquisition. A number of existing knowledge acquisition techniques are discussed. The existing techniques are based on shallow knowledge in terms of production rules. It is shown in this paper that process systems require a more systematic method of analysis, wherein a deep understanding of the physics of the process has to be integrated with heuristic information that needs to be extracted from operators or engineers. The process of modeling deep knowledge is achieved by the goal tree-success tree concept. The heuristic information is extracted from the answers of operators to simple questions made by the computer. The questions addressed to the operator are generated from the deep knowledge of the plant. The MAKE process discussed in this paper models the deep understanding and the heuristic information about a plant.