KNOB: a mutation-based approach for generating stereotype behaviors in active decision support systems

摘要

Role playing is a natural paradigm for decision support. It closely mimics the kind of support decision makers can expect to receive from their peers and fellow human beings. People, when called for decision support, often resort to the use of behavior patterns such as expert, devil's advocate, critique, playing dumb, careful listener, etc. These forms of support have a much higher potential for enhancing decision quality than the mundane forms of support such as number crunching, data analysis, etc. It is shown how knowledge base mutation may be used for generating such behavior patterns during decision support. The author covers the basic ideas of mutation, develops a systematic approach for applying the ideas for decision support, and illustrates its application for building a practical decision support system. The preliminary evaluations of the system are very encouraging, and suggest that systems incorporating these ideas can be very effective in practice.