Reasoning

 

Still missing : Case-based reasoning, demodulation+paramodulation, automatic programming.

• [Genesereth and Nilsson1987] Logical Foundations of Artificial Intelligence.
  Chapters 3, 4, 5. (Skip section 5.8 and the proofs in 4.10.)
  Logical inference, resolution and resolution strategies. (*) Should find something about natural theorem proving.
• [Reiter1987] Nonmonotonic Reasoning.
  A good technical overview of the field of nonmonotonic reasoning (*), but very outdated. I should write a summary of Makinson's paper, which should complement that.
• [Russel and Norvig1995] (*) Artificial Intelligence, a Modern Approach Chapters 14-16 deal with probability theory, Bayesian networks (other Uncertain reasoning approaches are mentioned briefly), utility theory and decision networks.
• [Genesereth and Nilsson1987] (*)
• [Pearl1990] Reasoning Under Uncertainty.
  A good review on uncertainty, includes Dempster Schafer. Ignore the details.
• [Charniak1991] (*)
• [Rich and Knight1991] Artificial Intelligence, Sections 7.5-7 The basic functionality of TMSs and ATMSs is described. Although Rich focuses on the nonmonotonic aspects of the mechanism, a TMS can also be used to cache inferences for a problem solver. (The interested reader might also read sections 7.2-4 on nonmonotonic reasoning ( (*) section 2 is pretty vage, but 3 and 4 give a nice view of the way nonmonotonic reasoning relates to other parts of AI)
• [de Kleer1986] An assumption-based TMS.
  Read section 1 to understand the basic ideas. The ATMS is a classic tradeoff that uses space (for storing contexts) to save time (needed to backtrack). Note that in contrast to Doyle's TMS, the ATMS is completely monotonic. (*) The reader should notice that s/he loses some of the ideas given in Doyle's TMS article, if s/he is restricted to the ATMS article.
• [Barr et al. 1989] Handbook of AI volume 4 Chapter 14, Blackboard systems by Penny Nii.
  An overview of the blackboard model.
• [Levesque and Brachman1985] A Fundamental Tradeoff in Knowledge Representation and Reasoning".
  There is a tradeoff between the expressiveness and the tractability of a Knowledge Representation scheme.

• [Levesque and Brachman1987] Expressiveness and Tractability in Knowledge Representation and Reasoning. Terminological languages provide a tradeoff between tractability and expressibility.
• [Baker and Shoham1992] Nonmonotonic Temporal Reasoning.
  Explanations of the frame, ramification, and qualification problems. Solutions to the Yale Shooting Problem. You may wish to follow up on the reference to Baker's solution which is considered the ``best one.'' (*) Should be replaced with something....
• [Cohen and Feigenbaum1982] The Handbook of Artificial Intelligence, Volume 3, Chapter 12.
  Skim section A for an overview of deduction. Read sections C (non-resolution theorem proving) and D (the Boyer-Moore theorem prover) (*)