Nayak, P.P. and Joskowicz, L. Efficient compositional modeling for generating causal explanations. Artificial Intelligence, 83:193--227, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....symbolic labels that characterize the phenomena it represents and its level of detail. To represent the relationships among model fragments, assumptions are organized into assumption classes ; the assumptions in each class represent mutually incompatible modeling alternatives. Several researchers [23, 38, 39] define interesting variants of this compositional modeling framework, but the basic ideas are the same. Our representation differs in two ways. First, the person encoding the domain knowledge need not group influences into model fragments; rather, in our approach, individual influences are the ....

....adequate model of a complex system, the number of elements that would have to be removed from the most detailed model is far greater than the number of elements that would have to be added to an empty model. For this reason, tripel takes the latter approach. Recent work by Nayak and Joskowicz [39] addresses the impracticality of simplifying a most detailed model. Their method generates an initial, overly detailed model and then applies Nayak s algorithm [38] to repeatedly simplify it. Their hope is that the initial model will be far simpler than the most detailed model. Their method ....

P. Pandurang Nayak and Leo Joskowicz. Efficient compositional modeling for generating causal explanations. Artificial Intelligence, 83:193--227, 1996.

....In the framework presented here, C t and C correspond to different behavioral specifications associated with models M t and M which are related via a focusing operation. Finally, this work intersects with work that offers a framework for building models from a library of model fragments (Nayak, [11], Falkenheimer and Forbus, 7] The difference between this work and that presented here is a difference in emphasis: in model building, the emphasis is on the construction of models out of sets of partial object models, whereas here the emphasis is on updating or modifying complete models ....

Nayak, P., and Joskowicz, L. Efficient Compositional Modeling for Generating Causal Explanations. Artificial Intelligence 83, 2, (1996), 193-228.

....a model building process is compositional modeling as in [ 6; 17; 14 ] and in the modeling algorithm of TRIPEL [ 21 ] This process can be followed by a model reformulation process. Model reformulation may involve structural consolidation [ 26 ] time scale selection [ 21 ] or simplification [ 18 ] or expansion of a model through addition, deletion, or replacement of a description of components or phenomena. Equation processing. Equation building produces an equation model 1 either directly from a task description or through reformulation of a model into a set of mathematical equations ....

....are mostly summarized in prose rather than exact and detailed definitions of relevant evaluators. The authors of the papers we discuss seldom provide a comparison of their techniques with other procedures, possibly because none exists. 4. 1 Model reformulation: Simplification Nayak and Joskowicz [ 18 ] propose a model reformulation technique that simplifies a compositional model of a device, while maintaining its ability to provide a causal explanation of the expected behavior of the device. The primary objective of their work is to perform efficient compositional modeling for generating ....

P. Pandurang Nayak and Leo Joskowicz. Efficient Compositional Modeling for Generating Causal Explanations. Artificial Intelligence, 83:193--227, 1996.

....quantity. The intuition underlying this algorithm is similar to the intuition underlying the construction of the query tree described in [21] which represents all the possible derivations of a query, and to other work on model formulation that follows causal chains in order to build models (e.g. [25, 33, 27]) 4.2.2 Selecting the Level of Detail The second part of the model selection problem is determining the level of detail at which to model each phenomenon. This entails deciding which abstractions and approximations can be made in modeling the system. Levy in [17, 16] demonstrates that knowledge ....

Pandurang Nayak and Leo Joskowicz. Efficient compositional modeling for generating causal explanations. Artificial Intelligence, 83:193--227, 1996.

....a model building process is compositional modeling as in [ 4; 15; 12 ] and in the modeling algorithm of TRIPEL [ 19 ] This process can be followed by a model reformulation process. Model reformulation may involve structural consolidation [ 24 ] time scale selection [ 19 ] or simplification [ 16 ] or expansion of a model through addition, deletion, or replacement of a description of components or phenomena. Equation processing. Equation building produces an equation model 1 either directly from a task description or through reformulation of a model into a set of mathematical equations ....

....evaluators. Moreover, the authors do not provide a comparison of their techniques with other procedures, possibly because none exists. Thus, we will not discuss comparators of problems and strategies in the context of these examples. 4. 1 Model reformulation: Simplification Nayak and Joskowicz [ 16 ] propose a model reformulation technique that simplifies a compositional model of a device, while maintaining its ability to provide a causal explanation of the expected behavior of the device. The primary objective of their work is to perform efficient compositional modeling for generating ....

P. Pandurang Nayak and Leo Joskowicz. Efficient Compositional Modeling for Generating Causal Explanations. Artificial Intelligence, 83:193--227, 1996.

....a model building process is compositional modeling as in [ 4; 15; 12 ] and in the modeling algorithm of TRIPEL [ 19 ] This process can be followed by a model reformulation process. Model reformulation may involve structural consolidation [ 23 ] time scale selection [ 19 ] or simplification [ 16 ] or expansion of a model through addition, deletion, or replacement of a description of components or phenomena. Equation processing. Equation building produces an equation model 1 either directly from a task description or through reformulation of a model into a set of mathematical equations ....

....evaluators. Moreover, the authors do not provide a comparison of their techniques with other procedures, possibly because none exists. Thus, we will not discuss comparators of problems and strategies in the context of these examples. 4. 1 Model reformulation: Simplification Nayak and Joskowicz [ 16 ] propose a model reformulation technique that simplifies a compositional model of a device, while maintaining its ability to provide a causal explanation of the expected behavior of the device. The primary objective of their work is to perform efficient compositional modeling for generating ....

P. Pandurang Nayak and Leo Joskowicz. Efficient Compositional Modeling for Generating Causal Explanations. Artificial Intelligence, 83:193--227, 1996.

....a model of both T base and S base , and hence (M base ) is a model of both T abs and S abs , and hence a model of T abs [ S abs . 2 Compositionality is exploited in diagnosis with multiple theories [ Nayak, 1994b ] and in compositional modeling [ Falkenhainer and Forbus, 1991; Nayak, 1994a; Nayak and Joskowicz, 1996; Iwasaki and Levy, 1994 ] where theories are built by composing knowledge from different sources. Moreover, one may argue that independent of these applications, theory compositionality is intrinsic to the very notion of abstractions. When T abs is an MI abstraction of T base , T abs can have ....

Nayak, P. P. and Joskowicz, L. Efficient Compositional Modeling for Generating Causal Explanations. Artif. Intell. To appear.

No context found.

Nayak, P.P. and Joskowicz, L. Efficient compositional modeling for generating causal explanations. Artificial Intelligence, 83:193--227, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC