Case-Based Reasoning Research and Development

We present an extension of an existing system, called SaxEx, capable of generating expressive musical performances based on Case-Based Reasoning (CBR) techniques. The previous version of SaxEx did not take into account the possibility of using affective labels to guide the CBR task. This paper discusses the introduction of such affective knowledge to improve the retrieval capabilities of the system. Three affective dimensions are considered—tender-aggressive, sad-joyful, and calm-restless that allow the user to declaratively instruct the system to perform according to any combination of five qualitative values along these three dimensions.

This paper is focused on a class of metrics for the Nearest Neighbor classifier, whose definition is based on statistics computed on the case base. We show that these metrics basically rely on a probability estimation phase. In particular, we reconsider a metric proposed in the 80’s by Short and Fukunaga, we extend its definition to an input space that includes categorical features and we evaluate empirically its performance. Moreover, we present an original probability based metric, called Minimum Risk Metric (MRM), i.e. a metric for classification tasks that exploits estimates of the posterior probabilities. MRM is optimal, in the sense that it optimizes the finite misclassification risk, whereas the Short and Fukunaga Metric minimize the difference between finite risk and asymptotic risk. An experimental comparison of MRM with the Short and Fukunaga Metric, the Value Difference Metric, and Euclidean-Hamming metrics on benchmark datasets shows that MRM outperforms the other metrics. MRM performs comparably to the Bayes Classifier based on the same probability estimates. The results suggest that MRM can be useful in case-based applications where the retrieval of a nearest neighbor is required.

Knowledge of the preferences of individual users is essential for intelligent systems whose performance is tailored for individual users, such as agents that interact with human users, instructional environments, and learning apprentice systems. Various memory-based, instance-based, and case-based systems have been developed for preference modeling, but these system have generally not addressed the task of selecting examples to use as queries to the user. This paper describes UGAMA, an approach to learning preference criteria through active exploration. Under this approach, Unit Gradient Approximations (UGAs) of the underlying quality function are obtained at a set of reference points through a series of queries to the user. Equivalence sets of UGAs are then merged and aligned (MA) with the apparent boundaries between linear regions. In an empirical evaluation with artificial data, use of UGAs as training data for an instance-based ranking algorithm (1ARC) led to more accurate ranking than training with random instances, and use of UGAMA led to greater ranking accuracy than UGAs alone.

Stratified case-based reasoning (SCBR) is a technique in which case abstractions are used to assist case retrieval, matching, and adaptation. Previous work has shown that SCBR can significantly decrease the computational expense required for retrieval, matching, and adaptation under a variety of different problem conditions. This paper extends this work to two new domains: a problem in combinatorial optimization, sorting by prefix reversal; and logistics planning. An empirical evaluation in the prefix-reversal problem showed that SCBR reduced search cost, but severely degraded solution quality. By contrast, in logistics planning, use of SCBR as an indexing mechanism led to faster solution times and permitted more problems to be solved than either hierarchical problem solving (by ALPINE) or ground level CBR (by SPA) alone. The primary factor responsible for the difference in SCBR’s performance in these two domains appeared to be that the optimal-case utility was low in the prefix-reversal task but high in logistics planning.

Since assigning indicies to textual cases is very time-consuming and can impede the development of CBR systems, methods to automate the task are desirable. In this paper,we present amachine learning approach that helps to bootstrap the development of a larger case base from a small collection of marked-up case summaries. It uses the marked-up sentences as training examples to induce a classifier that labels incoming cases whether an indexing concept applies. We illustrate how domain knowledge and linguistic information can be integrated with amachine learning algorithm to improve performance.The paper presents experimental resultswhich indicate the usefulness of learning from sentences and adding a thesaurus.We also consider the chancesand limitations of leveraging the learned classifiers for full-text documents.

Most traditional CBR systems are passive in nature, adopting an advisor role in which a user manually consults the system. In this paper, we propose a system architecture and algorithm for transforming a passive interactive CBR system into an active, autonomous CBR system. Our approach is based on the idea that cases in a CBR system can be used to model hypotheses in a situation assessment application, where case attributes can be considered as questions or information tasks to be performed on multiple information sources. Under this model, we can use the CBR system to continually generate tasks that are planned for and executed based on information sources such as databases, the Internet or the user herself. The advantage of the system is that the majority of trivial or repeated questions to information sources can be done autonomously through information gathering techniques, and human users are only asked a small number of necessary questions by the system. We demonstrate the application of our approach to an equipment diagnosis domain. We show that the system integrates CBR retrieval with hierarchical query planning, optimization and execution.

This paper presents the Case-Based Reasoning Real-Time Scheduling System (CBR-RTS) that integrates into a case-based reasoning framework a heuristic search component. The problem addressed involves scheduling sets of tasks with precedence, ready time and deadline constraints. CBR-RTS reuses the solution of known cases to simplify and solve new problems. When the system does not have applicable cases, it tries to find a solution using heuristic search. A particularly interesting feature of CBR-RTS is its learning ability. New problems solved by the heuristic scheduler can be added to the case base for future reuse. Performed tests have shown that small bases of cases carefully chosen allow to substantially reduce the time needed to solve new complex problems

Case-based reasoning exploits memorized problem solving episodes, called cases, in order to solve a new problem. Adaptation is a complex and crucial step of case-based reasoning which is generally studied in the restricted framework of an application domain. This article proposes a first analysis of case adaptation independently from a specific application domain. It proposes to combine the retrieval and adaptation steps in a unique planning process that builds an ordered sequence of operations starting from an initial state (the stated problem) and leading to a final state (the problem once solved). Thus, the issue of case adaptation can be addressed by studying the issue of plan adaptation. Finally, it is shown how case retrieval and case adaptation can be related thanks to reformulations and similarity paths.

The adaptation step is central in case-based reasoning (CBR), because it conditions the obtaining of a solution to a problem. This step is difficult from the knowledge acquisition and engineering points of view. We propose a knowledge level analysis of the adaptation step in CBR using the reasoning task concept. Our proposal is based on the study of several CBR systems for complex applications which imply the adaptation task. Three of them are presented to illustrate our analysis. We sketch from this study a generic model of the adaptation process using the task concept. This model is in conformity with other CBR formal models.

Current Case-Based Reasoning (CBR) process models present CBR as a low maintenance AI-technology and do not take the processes that have to be enacted during system development and utilization into account. Since a CBR system can only be useful if it is integrated into an organizational structure and used by more than one user, processes for continuous knowledge acquisition, -utilization and -maintenance have to be put in place. In this paper the short-comings of classical CBR process models are analyzed, and, based on the experiences made during the development of the case-based help-desk support system HOMER, the managerial, organizational and technical processes related to the development and utilization of CBR systems described.

In this paper Description Logics are presented as a suitable formalism to model the CBR life cycle. We propose a general model to structure the knowledge needed in a CBR system, where adaptation knowledge is explicitly represented. Next, the CBR processes are described based on this model and the CBR system OoFRA is presented as an example of our approach.

We present a case adaptation method that employs ideas from the field of genetic algorithms. Two types of adaptations, case combination and case mutation, are used to evolve variations on the contents of retrieved cases until a satisfactory solution is found for a new specified problem. A solution is satisfactory if it matches the specified requirements and does not violate any constraints imposed by the domain of applicability. We have implemented our ideas in a computational system called GENCAD, applied to the layout design of residences such that they conform to the principles of feng shui, the Chinese art of placement. This implementation allows us to evaluate the use of GA’s for case adaptation in CBR. Experimental results show the role of representation and constraints.

For the improvement of software quality and productivity, organizations need to systematically build up and reuse software engineering know-how, promoting organizational learning in software development. Therefore, an integrated support platform has to be developed for capturing, storing and retrieving software engineering knowledge. Technical support is complicated through specific characteristics of the software engineering domain, such as the lack of explicit domain models in practice and the diversity of environments. Applying Case-Based Reasoning, we propose an approach for the representation of relevant software engineering experiences, the goal-oriented and similarity-based retrieval tailorable to organization-specific characteristics and the continuous acquisition of new experiences. The approach is applied and validated in the context of the Goal/Question/Metric (GQM) approach, an innovative technology for software measurement.

Memory-based reasoning systems are a class of reasoners that derive solutions to new problems based on past experiences. Such reasoners use a long-term memory (LTM) to act as a knowledge base of these past experiences, which may be represented by such things as specific events (i.e. cases), plans, scripts, etc. This paper describes a Unified Long-Term Memory (ULTM) system, which is a dynamic, conceptual memory that was designed to be a general LTM capable of simultaneously supporting multiple intentional reasoning systems. Through a unique mixture of content-independent and domain-specific mechanisms, the ULTM is able to flexibly provide reasoners accurate and timely storage and recall of episodic memory structures. In addition, the ULTM provides support for recognizing opportunities to satisfy suspended goals, allowing reasoning systems to better cope with the unpredictability of dynamic real-world domains by helping them take advantage of unexpected events.

Because of the complexity of aerospace design, intelligent systems to support and amplify the abilities of aerospace designers have the potential for profound impact on the speed and reliability of design generation. This article describes a framework for supporting the interactive capture of design cases and their application to new problems, illustrating the approach with a discussion of its use in a support system for aircraft design. The project integrates case-based reasoning with interactive tools for capturing expert design knowledge through “concept mapping.” Concept mapping tools provide crucial functions for interactively generating and examining design cases and navigating their hierarchical structure, while CBR techniques provide capabilities to facilitate retrieval and to aid interactive adaptation of designs. The project aims simultaneously to develop a useful design aid and more generally to develop practical interactive approaches to fundamental issues of case acquisition and representation, context-sensitive retrieval, and case adaptation.

Case-based problem-solving systems reason and learn from experiences, building up case libraries of problems and solutions to guide future reasoning. The expected benefits of this learning process depend on two types of regularity: (1) problem-solution regularity, the relationship between problem-to-problem and solution-to-solution similarity measures that assures that solutions to similar prior problems are a useful starting point for solving similar current problems, and (2) problem-distribution regularity, the relationship between old and new problems that assures that the case library will contain cases similar to the new problems it encounters. Unfortunately, these types of regularity are not assured. Even in contexts for which initial regularity is sufficient, problems may arise if a system’s users, tasks, or external environment change over time. This paper defines criteria for assessing the two types of regularity, discusses how the definitions may be used to assess the need for case-base maintenance, and suggests maintenance approaches for responding to those needs. In particular, it discusses the role of analysis of performance over time in responding to environmental changes.

The case-based reasoning paradigm is applied in support of decision making processes related to pile foundations. Based on this paradigm, the system accumulates experience from previously realized pile foundations. This experience can be drawn when new situations with similar attributes of geotechnical situation of the site and geometric characteristics of the piles are encountered. Two case libraries were created based on previously realized sites. The representativeness of the case libraries and the efficiency of the search process are facilitated by the use of a genetic algorithm reduction.

As part of our investigation of how abstract principles are operationalized to facilitate their application to specific fact situations, we have begun to develop and experiment with SIROCCO (System for Intelligent Retrieval of Operationalized Cases and COdes), a CBR retrieval and analysis system applied to the domain of engineering ethics. SIROCCO is intended to retrieve decided engineering ethics cases and previously applied ethics codes to assist engineers and students in analyzing new cases. Here we describe a limited but expressive language designed to represent a wide range of ethics cases in SIROCCO, a world-wide web tool developed to perform case acquisition and support a measure of consistency in representation, and an experiment to validate the initial phase of SIROCCO’s retrieval algorithm and test its sensitivity to small variations in case description.

We discuss several problems of analogy-driven proof plan construction which prevent a solution for more difficult target problems or make a solution very expensive. Some of these problems are due to the previously assumed fixed order of matching, reformulation, and replay in case-based reasoning and from a too restricted combination of planning from first principles with the analogy process. In order to overcome these problems we suggest to interleave matching and replay as well as casebased planning with planning from first principles.Secondly, the restricted mixture of case-based planning and planning from first principles in previous systems is generalised to intelligently employing different planning strategies with the objective to solve more problems at all and to solve problems more efficiently.

This paper presents a policy to retain new cases based on retrieval benefits for case-based planning (CBP). After each case-based problem solving episode, an analysis of the adaptation effort is made to evaluate the guidance provided by the retrieved cases. If the guidance is determined to be detrimental, the obtained solution is retain as a new case in the case base. Otherwise, if the retrieval is beneficial, the case base remains unchanged. We will observe that the notion of adaptable cases is not adequate to address the competence of a case base in the context of CBP. Instead, we claim that the notion of detrimental retrieval is more adequate. We compare our retain policy against two policies in the CBP literature and claim that our policy to retain cases based on the benefits is more effective. Our claim is supported by empirical validation.

This paper describes HICAP, a general-purpose, interactive case-based plan authoring architecture that can be applied to decision support tasks to yield a hierarchical course of action. It integrates a hierarchical task editor with a conversational case-based planner. HICAP maintains both a task hierarchy representing guidelines that constrain the final plan and the hierarchical social organization responsible for these tasks. It also supports bookkeeping, which is crucial for real-world large-scale planning tasks. By selecting tasks corresponding to the hierarchy’s leaf nodes, users can activate the conversational case-based planner to interactively refine guideline tasks into a concrete plan. Thus, HICAP can be used to generate context sensitive plans and should be useful for assisting with planning complex tasks such as noncombatant evacuation operations. We describe an experiment with a highly detailed military simulator to investigate this claim. The results show that plans generated by HICAP were superior to those generated by alternative approaches.

The paper discusses the different aspects concerning performance arising in multi-modal systems combining Case-Based Reasoning and Model-Based Reasoning for diagnostic problem solving. In particular, we examine the relation among speed-up of problems solving, competence of the system and quality of produced solutions. Because of the well-know utility problem, there is no general strategy for improving all these parameters at the same time, so the trade-off among such parameters must be carefully analyzed. We have developed a case memory management strategy which allows the interleaving of learning of new cases with forgetting phases, where useless and potentially dangerous cases are identified and removed. This strategy, combined with a suitable tuning on the precision required for the retrieval of cases (in terms of estimated adaptation cost), provides an effective mechanism for taking under control the utility problem. Experimental analysis performed on a real-world domain shows in fact that improvements over both speed-up and competence can be obtained, without compromising in a significant way the quality of solutions.

The presented methods allow the modeling of student skills by a history of how well a student has performed on a series of tutoring cases. By this means, the explicit representation of a student’s knowledge, which is especially difficult for case oriented systems, can be avoided. Moreover, case histories make it possible to model the improvement of student’s skills and thus can be used for retrieving and adapting appropriate tutoring plans for further tutoring cases.On the other hand, the acquisition of tutoring cases is made easier by the methods described. The author of a tutoring case builds a set of test configurations for each association step that has to be performed within the tutoring case. The elements of the obtained sets just have to be ordered according to their difficulty. An explicit assignment of user stereotypes or difficulty levels to association tests is not necessary. Only a small set of model configurations for given student stereotypes must be defined.Further cooperative work within the scope of the project Docs’n Drugs will be dedicated to the implementation of thread configurations as corresponding networks of information and decision for testing and evaluating the proposed methods.Additionally, research will focus on more sophisticated similarity measures between thread configurations and explicit weightings of threads that take their different importance for the tutoring process into account.

Case-based reasoning systems solve problems by reusing a corpus of previous problem solving experience stored as a case-base of individual problem solving cases. In this paper we describe a new technique for constructing compact competent case-bases. The technique is novel in its use of an explicit model of case competence. This allows cases to be selected on the basis of their individual competence contributions. An experimental study shows how this technique compares favorably to more traditional strategies across a range of standard data-sets.

The success of a case-based reasoning system depends critically on the performance of the retrieval algorithm used and, specifically, on its efficiency, competence, and quality characteristics. In this paper we describe a novel retrieval technique that is guided by a model of case competence and that, as a result, benefits from superior efficiency, competence and quality features.

In response to the occurrence of an air incident, controllers at one of the three Canadian Rescue Coordination Centers (RCC) must make a series of critical decisions on the appropriate procedures to follow. These procedures (called incident prosecution) include hypotheses formulation and information gathering, development of a plan for the search and rescue (SAR) missions and in the end, the generation of reports. We present in this paper the results of a project aimed at evaluating the applicability of CBR to help support incident prosecution in the RCC. We have identified three possible applications of CBR: Online help, real time support for situation assessment, and report generation. We present a brief description of the situation assessment agent system that we are implementing as a result of this study.

Reliable and effective maintenance support is a vital consideration for the management within today’s manufacturing environment. This paper discusses the development a maintenance system for the world largest robot welding facility. The developed system combines a case-based reasoning approach for diagnosis with context information, as electronic on-line manuals, linked using open hypermedia technology. The work discussed in this paper delivers not only a maintenance system for the robot stations under consideration, but also a design framework for developing maintenance systems for other similar applications.

The integration of rule-based and case-based reasoning is particularly useful in medical applications, where both general rules and specific patient cases are usually available. In the present paper we aim at presenting a decision support tool for Insulin Dependent Diabetes Mellitus management relying on such a kind of integration. This multi-modal reasoning system aims at providing physicians with a suitable solution to the problem of therapy planning by exploiting, in the most flexible way, the strengths of the two selected methods. In particular, the integration is pursued without considering one of the modality as the most prominent reasoning method, but exploiting complementarity in all possible ways. In fact, while rules provide suggestions on the basis of a situation detection mechanism that relies on structured prior knowledge, CBR may be used to specialize and dynamically adapt the rules on the basis of the patient’s characteristics and of the accumulated experience. On the other hand, if a particular patient class is not sufficiently covered by cases, the use of rules may be exploited to try to learn suitable situations, in order to improve the competence of the case-based component. Such a work will be integrated in the EU funded project T-IDDM architecture, and has been preliminary tested on a set of cases generated by a diabetic patient simulator.

In many fields dependant upon complex observation, the structuring, depiction and treatment of knowledge can be of great complexity. For example in Systematics, the scientific discipline that investigates bio-diversity, the descriptions of specimens are often highly structured (composite objects, taxonomic attributes), noisy (erroneous or unknown data), and polymorphous (variable or imprecise data). In this paper, we present IKBS, an Iterative Knowledge Base System for dealing with such complex phenomena. The originality of this system is to implement the scientific method in biology: experimenting (learning rules from examples) and testing (identifying new individuals, improving the initial model and descriptions). This methodology is applied in the following ways in IKBS: 1 - Knowledge is acquired through a descriptive model that suits the semantic demand of experts. 2 - Knowledge is processed with an algorithm derived from C4.5 in order to take into account structured knowledge introduced in the previous descriptive model of the domain. 3 - Knowledge is refined through the use of an iterative process to evaluate the robustness of the descriptive model and descriptions. The IKBS system is presented here as a life science application facilitating the identification of coral specimens of the family Pocilloporidæ.

A case-based reasoning system for diagnosing medical equipment, called ELSI, has been in use by the GE corporation since 1994. When a customer or field engineer calls the service center for help with a problem, the equipment’s error log is automatically downloaded. In ninety seconds or less, ELSI displays a sorted list of the best-matching logs in a case base of previous known problems, shows the fix, service notes, explains which sections of the log match, and which fixes each section predicts. This diagnostic information allows the service center engineer to recommend a temporary work-around or remote fixes to a customer, or helps a field engineer show up on site with the right parts the first time.

This paper describes a software tool called CALIBRE (Candidate Library Retrieval). The tool incorporates case-base reasoning to support the extraction of candidate lists for targeted marketing campaigns. The tool has been aimed at users in the marketing domain. This domain is characterised by very large databases containing many Terabytes of customer related information. Large systems such as these require careful management of the queries being submitted to optimise the use of processing and storage resources. The CBR approach encourages consistent best practice as well as cutting down on valuable negotiation time. An early prototype has been built and is currently used for experimental purposes.

. The development of an Image Processing (IP) application is a complex activity, which can be greatly alleviated by user-friendly graphical programming environments. Our major objective is to help IP experts reuse parts of their applications. A first work towards knowledge reuse has been to propose a suitable representation of the strategies of IP experts by means of IP plans (trees of tasks, methods and tools). This paper describes the CBR module of our interactive system for the development of IP plans. After a brief presentation of the overall architecture of the system and its other modules, we explain the distinction between an IP case and an IP plan, and give the selection criteria and functions that are used for similarity calculation. The core of the CBR module is a search/adaptation algorithm, whose main steps are detailed: retrieval of suitable cases, recursive adaptation of the selected one and memorization of new cases. The system’s implementation is presently completed; its functioning is described in a session showing the kind of assistance provided by the CBR module during the development of a new IP application.

This paper describes an approach to represent and reuse efficiently the underlying design concepts in the existing mechanisms in order to synthesize mechanisms for function-generation and motion-transmission. A notion of virtual function generator is introduced to conceptualize and represent all possible underlying design concepts in the existing mechanisms. The virtual function generators are extracted from the existing mechanisms and composed of one or more primitive mechanisms together with the involved functions. They serve as new conceptual building blocks in the conceptual synthesis of design alternatives. The whole design concept or sub-concepts of the mechanisms can be represented and reused efficiently by the notion of virtual function generator. New mechanisms are generated by extracting and combining the underlying design concepts via the virtual function generators. The capability of the proposed approach is illustrated with a design example.

Case Based Reasoning has found increasing application on the Internet as an assistant in Internet commerce stores and as a reasoning agent for online technical support. The strength of CBR in this area stems from its reuse of the knowledge base associated with a particular application, thus providing an ideal way to make personalised configuration or technical information available to the Internet user. Since case data may be one aspect of a company’s entire corporate knowledge system, it is important to integrate case data easily within a company’s IT infrastructure, using industry specific vocabulary. We suggest XML as the likely candidate to provide such integration. Some applications have already begun to use XML as a case representation language. We review these and present the idea of a standard case view in XML that can work with the vocabularies or namespaces being developed by specific industries. Earlier research has produced version 1.0 of a Case Based Mark-up Language which attempts to mark-up cases in XML to enable distributed computing. The drawbacks of this implementation are outlined in this paper as well as the developments in XML that allow us to produce an XML “View” of a company’s knowledge system. We will detail the benefits of our system for industry in general in terms of extensibility, ease of reuse and interoperability.

The development of successful case-based design aids depends both on the CBR processes themselves and on crucial questions of integrating the CBR system into the larger task context: how to make the CBR component provide information at the right time and in the right form, how to access relevant information from additional information sources to supplement the case library, how to capture information for use downstream and how to unobtrusively acquire new cases. This paper presents a set of design principles and techniques that integrate methods from CBR and information retrieval to address these questions. The paper illustrates their application through a case study of the Stamping Advisor, a tool to support feasibility analysis for stamped metal automotive parts.

This paper details the way case-based reasoning has been used to aid footwear designers in creating new designs while maximizing component reuse. A hybrid system was created which uses an object-oriented memory model, neural networks for retrieval and fuzzy feature vectors to augment the basic case-based reasoning model. One of the main tasks involved in the design of any case-based system is determining the features that make up a case and finding a way to index these cases in a case-base for efficient and correct retrieval. This paper looks at the components of this footwear design system and how the various elements join together to create a useful system, in particular, how the use of fuzzy feature vectors and neural networks can improve the indexing and retrieval steps in case-based systems.

Nowadays, the complexity involved in computer network fault management demands a great amount of information about the involved technologies and their associated problems. Besides, Trouble Ticket Systems have been used to store the occurred problems, actuating as an historical memory of the network. Thus, a correct approach to consolidate the network historic memory is the development of an expert system that takes in account the knowledge accumulated in the Trouble Ticket Systems to propose solutions for an average problem. This work presents a system that uses Case-Based Reasoning paradigm applied to a Trouble Ticket System to suggest solutions for a new problem occurred. This system aims to aid diagnosis and resolution stages of network management problems. Typical problems of this domain, the proposed solution and results reached with the developed prototype are described.

Image Segmentation is a crucial step if extracting information from a digital image. It is not easy to set up the segmentation parameter so that it fits best over the entire set of images, which should be segmented. In the paper, we propose a novel architecture for image segmentation method based on CBR, which can adapt to changing image qualities and environmental conditions. We describe the whole architecture, the methods used for the various components of the systems and show how it performs on medical images.

CASA (computer aided systems architecting) is a methodology and tool to support the design of complex technical systems. It combines approaches from systems and requirement engineering and AI. System design in CASA is requirement-driven and works by a hierarchical stepwise top-down refinement of designs and a hierarchical decision making process. One important task in CASA deals with reusability of existing design artifacts and is supported by case-based reasoning techniques. Based on given structural specifications and formal requirements, a search procedure finds the best inexact match in a design base and computes an estimated degree of fulfillment for requirements. The approach employs efficient graph matching and indexing scheme for case retrieval and structural similarities and has adapted usual similarity measures to compute degree of fulfillment of requirements. It has been show by different example projects that the developed methods can be of great practical assistance for a designer.

In this paper, we describe case-based techniques in a medical application. We have developed a prototype of an antibiotics therapy adviser within the ICONS project, where the main advantage of applying CBR techniques is to speed-up the process of computing advisable therapies. However, some adaptations do not really belong to the Case-Based Reasoning paradigm though information from former cases is considered. They deal with rather typical medical tasks, namely modifications due to information updates. In our incrementally working system we have attempted to solve the problem of the continuously increasing number of stored cases by generalising from specific single cases to more general prototypes and by subsequently erasing redundant cases. Here we present results of experiments with threshold settings for our prototype architecture. The results show that the chosen design, which has mainly been founded on experiences with diagnostic applications, is not only advantageous for this therapeutic task, but that it contains a slight drawback as well.

In the future digital TV will offer an unprecedented level of programme choice. We are told that this will lead to dramatic increases in viewer satisfaction as all viewing tastes are catered for all of the time. However, the reality may be somewhat different. We have not yet developed the tools to deal with this increased level of choice (for example, conventional TV guides will be virtually useless), and viewers will face a significant and frustrating information overload problem. This paper describes a solution in the form of the PTV system. PTV employs user profiling and information filtering techniques to generate web-based TV viewing guides that are personalised for the viewing preferences of individual users. The paper explains how PTV constructs graded user profiles to drive a hybrid recommendation technique, combining case-based and collaborative information filtering methods. The results of an extensive empirical study to evaluate the quality of PTV’s casebased and collaborative filtering strategies are also described.

This paper describes a quality management system (called CBQM: Case-Based Quality Management) using the case-based reasoning mechanism which is based on a cost expectation value. The cost expectation value is calculated from objective and subjective values. We developed a quality management system that emplys a stochastic method. However, in some cases, this stochastic-based system failed to select good cases. Therefore, we have integrated some expectation values into the case selection mechanism. The CBQM has an expectation measurement. Its case selection criteria use not only similarity, but also some expectation values. If unforeseen malfunctions may occur due to inappropriate design, manufacturing condition and/or unsuitable usage, the similarity is not enough to select useful cases from a casebase. That is because the similarity is mainly based on products themselves. The CBQM adopts the cost expectation value in order to pick up useful cases. The CBQM’s selection criteria is based on the wuality of cases, which considers repair time, repair part cost, trouble recurrence, the confidence of diagnosis and repair difficulty. We validated this system in real product repair problems which field servive engineers repair home appliances.

Locomotives, like many modern complex machines, are equipped with the capability to generate on-board fault messages indicating the presence of anomalous conditions. Such messages tend to generate in large quantities and difficult and time consuming to interpret manually. This paper presents the design and development of a case-based reasoning system for diagnosing locomotive faults using such fault messages as input. The process of using historical repair data and expert input for case generation and validation is described. An algorithm for case matching is presented along with some results on pilot data.