Fox, S. & Leake D. (1994) Using Introspective Reasoning to Guide Index Refinement. Proceedings of the Sixteenth International Conference of the Cognitive Science Conference, (pp. 324 - 329)  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of the target design and hence have easilyadaptable matches. Bimbaum, Collins, Brand, Freed, Krulwich Pryor [3] proposed a system that learns to index cases on the basis of their adaptability, overriding semantic similarity where appropriate, a proposal that has been implemented by Fox Leake [4]. Their system avoids cases with feature combinations that were difficult to adapt in previous problem solving episodes. Leake s [5] method of constructive similarity also addresses the fundamental importance of adaptability in case retrieval. Cases matched is not viewed as from the viewpoint of ....

Fox, S. & Leake D. (1994) Using Introspective Reasoning to Guide Index Refinement. Proceedings of the Sixteenth International Conference of the Cognitive Science Conference, (pp. 324 - 329)

....of the target design and hence have easily adaptable matches. Birnbaum, Collins, Brand, Freed, Krulwich Pryor [3] proposed a system that learns to index cases on the basis of their adaptability, overriding semantic similarity where appropriate, a proposal that has been implemented by Fox Leake [4]. Their system avoids cases with feature combinations that were difficult to adapt in previous problem solving episodes. Leake s [5] method of constructive similarity also addresses the fundamental Presented as Invited Talk at: 2 nd UK Workshop on Case Based Reasoning, Manchester, UK. ....

Fox, S. & Leake D. (1994) Using Introspective Reasoning to Guide Index Refinement. Proceedings of the Sixteenth International Conference of the Cognitive Science Conference, (pp. 324 - 329)

....the conflict propagation potential of a case to assess ease of adaptation during retrieval. In KRITIK [5] candidate cases that satisfy certain functional specifications are preferred during retrieval, as those tend to have more easily adaptable matches. Birnbaum et al. 10] and Leake Fox [12] index cases on the basis of their adaptability, avoiding cases with feature combinations that were difficult to adapt in previous problem solving episodes. In our work, where several cases are being simultaneously combined, assessing adaptability of each individual case during retrieval is ....

David Leake and S. Fox. Using introspective reasoning to guide index refinement. In Proceedings of the Sixteenth International Conference of the Cognitive Science Society, pages 313--318, 1992.

....adaptability is assessed based on specially formulated adaptation knowledge, and is used to guide retrieval. Huang and Miles (Huang Miles 1995) calculate the conflict propagation potential of a case to assess ease of adaptation during retrieval. Birnbaum et al. L. Birnbaum 1989) and Leake Fox (Leake Fox 1992) index cases on the basis of their adaptability, avoiding cases with feature combinations that were difficult to adapt in previous problem solving episodes. Portinale et al. L. Portinale Magro 1997) describe a technique called Pivoting Based Retrieval, in the context of case based diagnosis, ....

Leake, D., and Fox, S. 1992. Using introspective reasoning to guide index refinement. In Proceedings of the Sixteenth International Conference of the Cognitive Science Society, 313--318.

....researchers have explored additional benefits. For instance, such knowledge can be used to facilitate the transfer of what is learned to other domains (Krulwich, 1991) A system can also use a self model of ideal system performance to determine when and how to refine memory retrieval criteria (Fox and Leake, 1994). Additional support for the use of self knowledge is found in the research of Jones (1991) Oehlmann et al. 1994) Owens (1990) and Stroulia and Goel (in press) In psychological studies of human metacognition (cognition about cognition) Wellman (1992) has also emphasized the role of ....

Fox, S., and Leake, D. (1994). Using Introspective Reasoning to Guide Index Refinement in Case-Based Reasoning.

....external world to achieve learning goals (e.g. access a database to answer a question) Meta AQUA s plans operate on the internal word of the system s background knowledge. Although many computational systems use a reflective reasoning approach (e.g. Collins, Birnbaum, Krulwich, Freed, 1993; Fox Leake, 1994; Oehlmann, Edwards, Sleeman, 1994; Plaza Arcos, 1993; Stroulia Goel, in press) and a few have used the planning metaphor in learning (Hunter, 1990; Quilici, in press; Ram Hunter, 1992; Ram Leake, in press; Redmond, 1992) none of these systems have applied the planning metaphor as ....

Fox, S., and Leake, D. (1994). Using Introspective Reasoning to Guide Index Refinement in Case-Based Reasoning. In Proceedings of the Sixteenth Annual Conferences of the Cognitive Science Society (pp. 324-329). Hillsdale, NJ: Lawrence Erlbaum.

....obtained solution. If no domain theory is available, the only way to evaluate the solution is to test it in the real world. Most systems modify the organization of the case base to take into account the occured failures : they update indexes in order to avoid future failures. Recent approaches ([2]) use an explicit model of the CBR process to diagnose reasoning failures by introspection. Michel Py ( 9] cites two reasoning strategies. We have added the third one. These strategies can be mixed in the CBR process to enhance the reasoning : ffl a dependency relations driven strategy : fi i ff ....

....not known, most CBR systems begin to retrieve previous cases by using a subset of problem descriptors of the new problem A 0 . In fact they are based on the assumption that these elements are relevant, i.e. somewhat the solution of the problem depends on these elements. The approach of S. Fox ([2]) is original because it tries to detect indexing failures after the adaptation step by using B 0 to search cases with a similar solution. The obtained cases are compared to the previously chosen one in order to validate and possibly refine indexing. ffl a matching driven strategy : ff A ; ff fi ....

Susan Fox and David B. Leake, `Using introspective reasoning to guide index refinement in case-based reasoning ', in Proceedings of the sixteenth annual conference of the cognitive science society, (1994).

....such scenario, as is the rice cook. Detecting the second class of failures is more di#cult; it requires expectations about the reasoning process itself, expectations about how long something should take, howmany steps should be involved, and how to tell if proper progress is being made #see also Fox Leake #1994##. Determining the ultimate cause: Once the system has detected an expectation failure, it must be able to determine what point in the reasoning process the detected failure was created. This task is perhaps the most di#cult part of introspective evaluation of the reasoning process, because ....

Fox, S. & Leake, D. #1994#. Using introspective reasoning to guide index re#nement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pp. 324#329 Atlanta, GA. Lawrence Erlbaum Associates.

....the problem down into individual actions of the simulated robot. 3 The Introspective Reasoning Task ROBBIE s high level task is to improve its reasoning process when reasoning failures are detected. ROBBIE s approach uses a declarative model describing the underlying system s reasoning processes (Fox Leake, 1994; Birnbaum et al. 1991; Freed Collins, 1994) The model contains assertions about the system s ideal reasoning behavior; the actual reasoning is compared to this ideal to detect discrepancies which are viewed as reasoning failures. Once a reasoning failure is detected the model is used to ....

....assertions about the system s ideal reasoning behavior; the actual reasoning is compared to this ideal to detect discrepancies which are viewed as reasoning failures. Once a reasoning failure is detected the model is used to create an explanation of the failure and to suggest a repair (See Fox Leake (1994) for a discussion of ROBBIE s failure detection) Currently, ROBBIE can re index its memory to include new features. The model s assertions describe states that would be ideally true of particular portions of the case based reasoning process. Assertions range from details of ROBBIE s ....

Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in casebased reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pp. 324--329 Atlanta, GA. Lawrence Erlbaum Associates.

....aspects of desired performance of the reasoning process as well for example, the allowable processing cost. Such a model provides a benchmark for evaluating the performance of the reasoning process (e.g. Birnbaum, Collins, Freed, Krulwich (1990) Collins, Birnbaum, Krulwich, Freed (1993) Fox Leake (1994, 1995) Both the models of expected and ideal behavior must be compared to the actual behavior of the reasoning process in reasoning episodes. It is clear that in an agent with perfect self understanding and perfect understanding of the external world, actual performance would always bear out ....

....search capabilities or of the contents of its memory. ffl Discrepancies between actual behavior and ideal behavior, whether or not these discrepancies are expected, show the need for learning to refine the processing of the reasoning system (e.g. Collins et al. 1993) Freed Collins (1994) Fox Leake (1994), Krulwich (1991) Ram Cox (1994) The success of this method for guiding learning depends on addressing two key problems. First, it is obvious that both the models of expected and ideal behavior will in general be incomplete or too abstract to apply to some sub parts of the reasoning process. ....

[Article contains additional citation context not shown here]

Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pp. 324--329 Atlanta, GA.

....a system that applies this process to refine the indices used for case retrieval in a case based planning system. Our program, ROBBIE, combines a case based planner as its performance system with an introspective reasoning component which monitors its reasoning and guides introspective learning [ Fox and Leake, 1995; 1994 ] Applying introspective reasoning to CBR is novel in that learning in CBR systems is generally focused on acquiring domain knowledge, by acquiring new cases. ROBBIE learns new cases, but its primary focus is on the use of introspective reasoning to refine how its cases are applied. The ....

....steps for the simulated robot to execute. 3.2 The Introspective Task The higher level task for ROBBIE is to improve its reasoning processes in response to detected reasoning failures. ROBBIE s approach uses an explicit, declarative model describing the underlying system s reasoning processes [ Fox and Leake, 1994; Birnbaum et al. 1991; Freed and Collins, 1994 ] The model provides expectations about the ideal reasoning behavior of the system; the actual reasoning of the system is compared to this ideal as a means of detecting reasoning failures. Once a reasoning failure is detected the model is used to ....

[Article contains additional citation context not shown here]

S. Fox and D. Leake. Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pages 324--329, Atlanta, GA, 1994. Lawrence Erlbaum Associates.

....reasoning process (in particular, its feature selection process) when it detects failures in its reasoning. As the planner reasons about its task, the introspective component monitors its reasoning process by comparing it to a declarative model describing the planner s ideal reasoning processes [6, 9, 12]. Reasoning failures occur when the model s expectations about the reasoning process are not fulfilled by the actual reasoning. Expectations are encoded as assertions; facts that would be ideally true of particular points in the case based reasoning process. For example, one assertion states ....

....Failure detection for the introspective reasoner involves noticing discrepancies between assertions and actual behavior. At each point in the reasoning process, the introspective reasoner compares the actual reasoning to the ideal. When they fail to match, a reasoning failure has occurred (See [9] for a more extensive discussion of ROBBIE s failure detection) Detecting failures due to faulty indexing criteria requires monitoring the entire reasoning process: the failure may not be discovered until after the inappropriately retrieved case has been adapted and executed. The problem is ....

[Article contains additional citation context not shown here]

S. Fox and D. Leake. Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pages 324--329, Atlanta, GA, 1994. Lawrence Erlbaum Associates.

....better use of their cases augmenting the domain expertise embodied in their libraries of problem solving cases with increased expertise at the case based reasoning process itself. This effort applies introspective reasoning and learning methods to improve memory organization retrieval criteria (Fox Leake, 1994, 1995) and case adaptation (Leake, 1993, 1994) After sketching some psychological perspective on case based reasoning as a model of expertise we summarize ongoing research on how a casebased reasoner can acquire expertise at case adaptation as a natural part of its case adaptation process. 2 ....

Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in case-based reasoning.

No context found.

University. Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society Atlanta, GA.

....scenario, as is the rice cook. Detecting the second class of failures is more difficult; it requires expectations about the reasoning process itself, expectations about how long something should take, how many steps should be involved, and how to tell if proper progress is being made (see also Fox Leake (1994)) Determining the ultimate cause: Once the system has detected an expectation failure, it must be able to determine what point in the reasoning process the detected failure was created. This task is perhaps the most difficult part of introspective evaluation of the reasoning process, because ....

Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pp. 324--329 Atlanta, GA. Lawrence Erlbaum Associates.

....similarity assessment, as well as at stored cases that are easier to adapt: ffl Refining indices to favor more adaptable cases Because the difficulty of case adaptation depends crucially on the cases that are retrieved, improvements in retrieval can significantly ameliorate the adaptation task. Fox and Leake (1994; 1995b; 1995c) apply introspective reasoning after problem solving to evaluate whether the best case was retrieved, and, if not, to adjust retrieval criteria to focus future retrievals on more adaptable cases. ffl Basing retrieval directly on adaptability Given that indexing and similarity ....

Fox, S., and Leake, D. 1994. Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, 324--329. Hillsdale, NJ: Lawrence Erlbaum.

.... Goel, 1994) uses an Structure BehaviorFunction model of its own reasoning to find learning opportunities, and IULIAN (Oehlmann, Edwards, Sleeman, 1994, 1995) uses questions about its own reasoning and knowledge to re index its memory and to regulate its processing. Our approach, ROBBIE 1 (Fox Leake, 1994), models the desired behavior of its underlying case based planning component as a set of expectations about the behavior of the system during the planning process. ROBBIE monitors the reasoning of its underlying system, comparing its performance to a model of the ideal performance of the ....

....specific to a particular implementation. The model structure preserves as much generality as possible by maintaining a hierarchy of assertions (expectations) which keep task and implementationspecific details separate from generalities that might be more transferable to other tasks and domains (Fox Leake, 1994). 1 Re Organization of Behavior By Introspective Evaluation Case based Reasoner Model based Planner World Simulator Monitor Real map Internal map Model of planner Repairs Memory Figure 1: ROBBIE Architecture Evaluating the method: Evaluation of AI systems is important to verify that the ....

[Article contains additional citation context not shown here]

Fox, S. & Leake, D. (1994). Using introspective reasoning to guide index refinement in case-based reasoning. In Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, pp. 324--329 Atlanta, GA.

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