M. Brown. A Memory-Model for Case Retrieval by Activation Passing. Ph.D Thesis (UMCS94 -2-1). University of Manchester, United Kingdom, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....selective retention. Selective utilization filters [Markovitch Scott, 1989] prevent the problem solver from seeing certain knowledge items for some period of time, and are often implemented in CBR systems as indexing schemes [Kolodner, 1994] or by time bounding the case retrieval mechanism [Brown, 1993; Veloso, 1992] In these approaches only a subset of the case base is made available to the retrieval mechanism with the attendant disadvantage that certain cases may be missed at retrieval time. Unfortunately, these omissions can result in unnecessarily complex adaptation stages or even ....

M. Brown. A Memory-Model for Case Retrieval by Activation Passing. Ph.D Thesis (UMCS94 -2-1). University of Manchester, United Kingdom, 1993.

....by product of reasoning (Section 1.1 of [15] This enables a CBR system to become more efficient and more competent as it gains in experience, as opposed to traditional expert systems which are brittle because they cannot improve beyond their particular plateaux of expertise (Section 1.2. 2 of [17]) 3.2 Why CBR Should Be Applied to Code Understanding and Generation Automatic programming in its full generality is an extremely difficult task to automate. Rich and Waters [8] suggest that the ultimate goal of automatic programming has three features: that it is fully automated; that it ....

Brown MG. A memory model for case retrieval by activation passing. PhD thesis, University of Manchester, Manchester, 1993, technical report UMCS-94-2-1. ftp:// ftp.cs.man.ac.uk/pub/TR/UMCS-94-2-1.ps.Z

.... new cases) which occurs as a natural by product of reasoning [8] This enables a CBR system to become more efficient and more competent as it gains in experience, as opposed to traditional expert systems which are brittle because they cannot improve beyond their particular plateaux of expertise [10]. Further advantages of CBR are that it alleviates the knowledge elicitation bottleneck [11] it is often easier, in complex domains, to elicit cases than a complete set of rules. A CBR system is often easier to build and maintain than the equivalent rule based system [8] which suggests that ....

Brown MG. A memory model for case retrieval by activation passing. PhD thesis, University of Manchester, Manchester, 1993. Technical report UMCS-94-2-1. Internet: ftp://ftp.cs.man.ac.uk/pub/TR/UMCS-94-2-1.ps.Z

....3.5 Conclusion Human experts are not systems of rules, they are libraries of experiences. Riesbeck and Schank (1989) Ruled based systems are often brittle because they can t learn from experience and they collapse dramatically when faced with a problem beyond their plateau of expertise [2]. On the other hand, case based reasoning can deal with such situations by acquiring new cases. The PTA agent needs to adapt itself to the user s behaviour but, because we don t know his behaviour in advance, it is not possible to create a set of rules adapted for each user. A case based reasoning ....

Brown M.G.: `A memory model for case retrieval by activation passing', Ph.D. dissertation, Department of Computer Science, University of Manchester, technical report UMCS-94-2-1, (1993)

....most similar to the current problem and hence are most promising for solving it. And, of course, retrieval should be efficient. Consequently, a number of sophisticated techniques have been developed, including kd trees ( 14, 15] the Fish and Sink approach ( 12] the Crash memory model ([4]) and Knowledge directed Spreading Activation (KDSA, 16] Recently, the model of Case Retrieval Nets (CRNs) has been introduced as an alternative memory model for the task of case retrieval ( 10, 9] CRNs show to be particularly suitable for analytic CBR tasks, such as classification, ....

....of an arbitrary case and applying the obtained result to other cases near the original selected one. The major advantage is that a cheap to compute aspect distance can be used to estimate the usefulness of cases and to avoid the (expensive) computation of the view distance. The Crash system ([4, 5]) uses an activation passing scheme for case retrieval, too. The major difference to the model of CRNs is that during similarity propagation activation is propagated through a network of world knowledge. On the one hand, this allows for the integration of a deep background knowledge into the case ....

M. G. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, University of Manchester, 1994.

....of techniques allowing for an efficient and yet flexible retrieval of relevant cases. This has led to a number of sophisticated techniques for this subtask, as for example indexing techniques ( 13] kd trees ( 17, 18] the heuristic Fish and Sink approach ( 14] the Crash memory model ([3]) and Knowledge directed Spreading Activation (KDSA, 19] As an alternative technique especially suitable for the design of decision support systems ( 4, 8] we developed the concept of Case Retrieval Nets (CRNs) as a memory structure supporting efficient but nevertheless flexible case ....

....The complete documentation of the results can be found in [7] Before we go into the details about CRNs, we would like to outline what we consider as requirements for a case retrieval technique. Basically, there are three conditions to be met: efficiency, completeness, and flexibility (cf. also [3, 17]) Efficiency concerns the effort required to access relevant cases. Access of these cases should avoid exhaustive search in memory. Completeness assures that every sufficiently similar case in memory will be found during retrieval. The corresponding problem of correctness is often solved as a ....

[Article contains additional citation context not shown here]

M. G. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, University of Manchester, 1994.

....of techniques allowing for an efficient retrieval of relevant cases in a given problem situation. This has led to a number of sophisticated techniques for this subtask, as for example indexing techniques ( 11] kd trees ( 14, 15] the Fish and Sink approach ( 12] the Crash memory model ([3]) and Knowledge directed Spreading Activation (KDSA, 16] As an alternative memory model we have developed the model of Case Retrieval Nets (CRNs [9] which are particularly suitable for analytic CBR tasks, such as classification, diagnosis, and decision support ( 6, 8] The goal of this ....

....of cases and to avoid the (expensive) computation of the view distance. Q Steps 1 2 Step 3 Step 4 Steps 5 6 Step 7: Presentation to the user All useful symptoms Step 7: Most Useful symptoms Step 7: Fig. 5. Process of differential diagnosis in the example OCRN of Figure 3 The Crash system ([3, 4]) uses an activation passing scheme for case retrieval, too. The major difference to (B)CRNs is that during similarity propagation activation is propagated through a network of world knowledge. On the one hand, this allows for the integration of a deep background knowledge into the case retrieval ....

M. G. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, University of Manchester, 1994.

....has mostly been investigated for representational approaches (using spreading activation and or parallel access to the case base through more than one index, e.g. as in the PARADYME system [10] even though these are already optimised towards their given similarity measure. Indeed, Brown [3] argues that further indexing of the case base may improve efficiency without resorting to parallel implementations. Parallelism for more computational approaches has been much less investigated (but see, e.g. 12] Similarity measures have traditionally been constructed on a one off basis by ....

M. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, Department of Computer Science, University of Manchester, 1994. Technical Report 94-2-1.

....techniques allowing for an efficient and yet flexible retrieval of relevant cases. This research has led to a number of sophisticated techniques for this subtask, as for example indexing techniques ( 8] kd trees ( 12, 13] the heuristic Fish and Sink approach ( 9] the Crash memory model ([1]) and Knowledge directed Spreading Activation (KDSA, 14] In an earlier paper ( 6] we presented ABS, a summary of our initial ideas of how to structure the case memory in order to support efficient case retrieval. In these ideas, however, some fundamental parts were still missing, such as a ....

....do not have a counterpart. Fish and Sink works by determining the utility of an arbitrary case and applying the obtained result to other cases near the original selected one. In order to assure completeness, one has, however, still to search the entire case base exhaustively. The Crash system ([1]) uses an activation passing scheme for case retrieval, too. The major difference to CRNs is that in step 2 of retrieval activation is propagated through a network of world knowledge. On the one hand, this allows for the integration of a deep background knowledge into the case retrieval system. On ....

M. G. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, University of Manchester, 1994.

.... N1 # (7 Gamma t) e = min fdepth e t 0 j t 0 2 (ae Tree t) # 2 Tree t 0 g Tree : Tree elem elem (elem elem bool) e1 ( Tree t #) e2 = # (7 Gamma t) e1 # (7 Gamma t) e2 Fig. 2. Similarity measure generating functions for trees found in many case based reasoning systems [2, 6]. There is, however, no reason why such an algorithm could not be used to define an ordering on graphs, and then apply this ordering in the way presented in this paper. Graphs. A similar construction can be used for graphs in general, by defining the distance function to return the length of the ....

M. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, Department of Computer Science, University of Manchester, 1994. Technical Report 94-2-1.

....that the relevant cases can be rapidly selected from large case bases. To satisfy this requirement, a number of techniques have been proposed recently (see [5] for an overview) including indexing schemes ( 11] kd trees ( 13] the Fish and Sink approach ( 12] the Crash memory model ([3]) and Case Retrieval Nets (CRNs) The latter employ net like memory structures to perform a retrieval similar to information access in an associative memory model. In a nutshell, cases are described as set of Information Entities (IEs) which represent single features of each case. These IEs are ....

M. G. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, University of Manchester, 1994.

....be found. Ironically, parallelism has mostly been investigated for representational approaches (using spreading activation and or parallel access to the case base through more than one index) even though these approaches are already optimised towards their given similarity measure. Indeed, Brown [Bro94] argues that further indexing of the case base may improve efficiency without resorting to parallel implementations. Parallelism for computational approaches has been much less investigated (but see, e.g. MT94] Existing computational CBR systems typically have very simple similarity ....

M. Brown. A Memory Model for Case Retrieval by Activation Passing. PhD thesis, Department of Computer Science, University of Manchester, 1994. Technical Report 94-2-1.

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