Reconstructive Explanation: Explanation as Complex Problem Solving

Exist ing explanation facilities are typical ly far more appropriate for knowledge engineers engaged in system maintenance than for endusers of the system. This is because the explanat ion is l i t t le more than a trace of the detailed problem-solving steps. An alternative approach recognizes that an effective explanat ion often needs to substantial ly reorganize the line of reasoning and br ing to bear addit ional informat ion to support the result. Explanat ion itself becomes a complex problem-solving process that depends not only on the line of reasoning, but also on addit ional knowledge of the domain. We present a new computat ional model of explanation and argue that it results in significant improvements over t radi t ional approaches.