Reasoning about assumptions in graphs of models.

*(English)*Zbl 0719.68066
IJCAI 89, Proc. Int. Conf., Detroit, MI/USA 1989, 1432-1438 (1989).

[For the entire collection see Zbl 0707.68001.]

It is well-known to be crucial for problem-solving paradigms based on models to include mechanisms for automatically and efficiently selecting a better model when the analysis in the current model is found to be in error. The authors represent physical domains as Graphs of Models (GM), where models are nodes of the graph and edges are the assumptions that have to be changed in going from one model to the other. The GM introduced paradigm includes methods that automatically change models when the current model is inadequate. This methods suppose to comprise (at least) four abilities: (1) to detect conflicts, (2) to determine how parameters must change in order to eliminate conflicts, (3) to represent how assumption transitions affect parameters in the world, and (4) to use this knowledge for selecting the next model. The paper introduces the notions of delta-vectors to capture the qualitative nature of parametric changes that will eliminate conflicts, and the parameter-change rules to capture domain-level knowledge about how assumption transitions affect values of parameters. A simple mechanism uses delta-vectors and parameter-change rules to decide which assumptions to change. Four implementations, albeit limited to domains containing 4-8 models, in geometric structure, thermodynamics, mechanics, and fluids, lead to the conclusion that the GM paradigm is a powerful approach to represent complex, scientific, and engineering domains. The qualitative mechanisms of delta-vectors and parameter-change rules provide efficient model changing behaviors.

It is well-known to be crucial for problem-solving paradigms based on models to include mechanisms for automatically and efficiently selecting a better model when the analysis in the current model is found to be in error. The authors represent physical domains as Graphs of Models (GM), where models are nodes of the graph and edges are the assumptions that have to be changed in going from one model to the other. The GM introduced paradigm includes methods that automatically change models when the current model is inadequate. This methods suppose to comprise (at least) four abilities: (1) to detect conflicts, (2) to determine how parameters must change in order to eliminate conflicts, (3) to represent how assumption transitions affect parameters in the world, and (4) to use this knowledge for selecting the next model. The paper introduces the notions of delta-vectors to capture the qualitative nature of parametric changes that will eliminate conflicts, and the parameter-change rules to capture domain-level knowledge about how assumption transitions affect values of parameters. A simple mechanism uses delta-vectors and parameter-change rules to decide which assumptions to change. Four implementations, albeit limited to domains containing 4-8 models, in geometric structure, thermodynamics, mechanics, and fluids, lead to the conclusion that the GM paradigm is a powerful approach to represent complex, scientific, and engineering domains. The qualitative mechanisms of delta-vectors and parameter-change rules provide efficient model changing behaviors.

Reviewer: N.Curteanu (Iaşi)

##### MSC:

68T20 | Problem solving in the context of artificial intelligence (heuristics, search strategies, etc.) |

68T15 | Theorem proving (deduction, resolution, etc.) (MSC2010) |

##### Software:

MACSYMA##### References:

[1] | Addanki, S.; Cremonini, R.; Penberthy, J. S.: Reasoning about assumptions in graphs of models. Proceedings IJCAI-89, 1432-1438 (1989) · Zbl 0719.68066 |

[2] | Bobrow, D. G.: Qualitative reasoning about physical systems. (1984) |

[3] | Works, Boston Gear: The Boston gear catalog. 59 (1960) |

[4] | Brown, J. S.; Burton, R. R.; De Kleer, J.: Pedagogical, natural language and knowledge engineering techniques in SOPHIE I, II, and III. Intelligent tutoring systems, 227-282 (1982) |

[5] | Burstall, A. F.: Simple working models of historic machines. (1968/1969) · Zbl 0213.43202 |

[6] | Candee, A. H.: Introduction to the kinematic geometry of gear teeth. (1961) |

[7] | Davis, R.: Diagnostic reasoning based on structure and behavior. Artif. intell. 24, 347-410 (1984) |

[8] | De Kleer, J.: Local methods of localizing faults in electronic circuits. Tech. report AIM-394 (1976) |

[9] | De Kleer, J.: An assumption-based TMS. Artif. intell. 28, 127-162 (1986) |

[10] | De Kleer, J.; Brown, J. S.: A qualitative physics based on confluences. Artif. intell. 24, 7-83 (1984) |

[11] | De Kleer, J.; Williams, B. C.: Reasoning about multiple faults. Proceedings AAAI-86, 132-139 (1986) |

[12] | Doyle, J.: A truth maintenance system. Artif. intell. 12, 231-272 (1979) |

[13] | Falkenhainer, B.; Forbus, K. D.: Compositional modeling: finding the right model for the job. Artif. intell. 51, 95-143 (1991) |

[14] | Forbus, K. D.: Qualitative process theory. Artif. intell. 24, 85-168 (1984) |

[15] | Forbus, K. D.; De Kleer, J.: Focusing the ATMS. Proceedings AAAI-88, 193-198 (1988) |

[16] | Hamscher, W. C.; Davis, R.: Diagnosing circuits with state: an inherently underconstrained problem. Proceedings AAAI-84, 142-147 (1984) |

[17] | Lowry, M. R.: The abstraction/implementation model of problem reformulation. Proceedings IJCAI-87, 1004-1010 (1987) |

[18] | Martin, W. A.; Fateman, R. J.: The MACSYMA system. Proceedings second ACM symposium on symbolic algebraic manipulation (1971) |

[19] | Mavrovouniotis, M.; Stephanopolous, G.: Reasoning with orders of magnitude and approximate relations. Proceedings AAAI-87, 626-630 (1987) |

[20] | Murthy, S.: Qualitative reasoning at multiple resolutions. Proceedings AAAI-88, 296-300 (1988) |

[21] | Murthy, S.; Addanki, S.: PROMPT: an innovative design tool. Proceedings AAAI-87, 432-637 (1987) |

[22] | Newell, A.; Simon, H. A.: Computer science as empirical enquiry: symbols and search. Commun. ACM 19, No. 3, 113-126 (1976) |

[23] | Penberthy, J. S.: Incremental analysis and the graph of models. S.M. thesis (1987) |

[24] | Raiman, O.: Order of magnitude reasoning. Proceeding AAAI-86, 100-104 (1986) |

[25] | Subramanian, D.; Genesereth, M. R.: The relevance of irrelevance. Proceedings IJCAI-87, 416-422 (1987) |

[26] | Weld, D.: Comparative analysis. Artif. intell. 36, 333-374 (1988) · Zbl 0654.93004 |

[27] | Weld, D.: Approximation reformulations. Proceedings AAAI-90, 407-412 (1990) |

This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.