Learning by analogical reasoning in general problem-solving

This dissertation integrates derivational analogy into general problem solving as a method of learning at the strategy level to solve problems more effectively. The derivational analogy method has been fully implemented in the sc PRODIGY architecture and proven empirically to be amenable to scaling up both in terms of domain and problem complexity. Reasoning by analogy involves a set of challenging problems, namely: how to accumulate episodic problem solving experience, how to define and decide when two problem solving situations are similar, how to organize large amounts of knowledge so that it may be efficiently retrieved, and finally the ultimate problem of how to successfully transfer chains of reasoning from past experience to new problem solving situations when only a partial match exists among corresponding problems. More specifically, the dissertation automates the generation, storage, dynamic indexation, retrieval and replay for multiple cases (i.e. derivational traces of past problem solving episodes). Learning occurs by accumulation and flexible reuse of cases. The problem solving search effort is reduced incrementally as more episodic experience is compiled into the case library. Scaling up the system proved to be very demanding. The current system has thus far been demonstrated in multiple domains, including a complex logistics transportation domain where it generated a library of 1000 cases, showed strong improvements in problem-solving performance, and pushed the solvability envelope to increasingly more complex classes of problems.