Organization and Inference in a Frame-Like System of Common Sense Knowledge

My goals have not changed since (Charniak 72). I am still interested in the construction of a computer program which will answer questions about simple narration (e.g. children's stories). More exactly, if one makes the somewhat unrealistic division of the problem into (a) going from natural language to a convenient internal representation, and (b) being able to "reason" about the information in the story in order to answer questions, my interests are clearly in the latter section. I will take it as given that such reasoning requires large amounts of "common sense knowledge" about the topics mentioned in the text, so I will not demonstrate this point. (However it should come out incidentally from the examples used to demonstrate other points.) To reason with this knowledge requires that it be organized, by which I simply mean it must be structured so that the system can get at necessary knowledge when it is needed, but that unnecessary knowledge will not clog the system with the all too familiar "combinatorial explosion". I will start with my current thoughts on organization.