Anders Kofod-Petersen, Agnar Aamodt: Contextualised ambient intelligence through case-based reasoning. To appear in Proceedings of the 8th European Conference on Case-Based Reasoning, ECCBR 2006, ldeniz/Fethiye, Turkey, September 4th-7th, 2006. Lecture Notes in Computer Science, Springer Verlag.
Ambient Intelligence is a research area that has gained a lot
of attention in recent years. One of the most important issues for ambient
intelligent systems is to perceive the environment and assess occurring
situations, thus allowing systems to behave intelligently. As the ambient
intelligence area has been largely technology driven, the abilities of systems
to understand their surroundings have largely been ignored. This
work demonstrates the first steps towards an ambient intelligent system,
which is able to appreciate the environment and reason about occurring
situations. This situation awareness is achieved through knowledge
intensive case-based reasoning.
Mingyang Gu, Agnar Aamodt: Evaluating CBR systems using different data sources: a case study. To appear in Proceedings of the 8th European Conference on Case-Based Reasoning, ECCBR 2006, ldeniz/Fethiye, Turkey, September 4th-7th, 2006. Lecture Notes in Computer Science, Springer Verlag.
The complexity and high construction cost of case bases make it very difficult, if not impossible, to evaluate a CBR system, especially a knowledge-intensive CBR system, using statistical evaluation methods on many case bases. In this paper, we propose an evaluation strategy, which uses both many simple case bases and a few complex case bases to evaluate a CBR system, and show how this strategy may satisfy different evaluation goals. The identified evaluation goals are classified into two categories: domain-independent and domain-dependent. For the evaluation goals in the first category, we apply the statistical evaluation method using many simple case bases (for example, UCI data sets); for evaluation goals in the second category, we apply different, relatively weak, evaluation methods on a few complex domain-specific case bases. We apply this combined evaluation strategy to evaluate our knowledge-intensive conversational CBR method as a case study.
Conversational Case-Based Reasoning (CCBR) provides a mixed-initiative dialog for guiding users to refine their problem descriptions incrementally through a question-answering sequence. In this paper, we argue that the successful dialogs in CCBR can be captured and learned in order to improve the efficiency of CCBR from the perspective of shortening the dialog length. A framework for dialog learning in CCBR is proposed in the present paper, and an instance of this framework is implemented and tested empirically in an attempt to evaluate the learning effectiveness of the framework. The results show us that on 29 out of the 32 selected datasets, CCBR with the dialog learning mechanism uses fewer dialog sessions to retrieve the correct case than CCBR without using dialog learning..
Ramon Lpez de Mntaras, David McSherry, Derek Bridge, David Leake, Barry Smyth, Susan Craw, Boi Faltings, Mary Lou Maher, Michael Cox, Kenneth Forbus, Mark Keane, Agnar Aamodt, and Ian Watson: Retrieval, reuse, revision, and retention in case-based reasoning . To appear in: Knowledge Engineering Review, Special Issue on Case-Based Reasoning. 2006. 32 pgs.
Case-based reasoning (CBR) is an approach to problem solving that emphasizes the role of prior experience during future problem solving (i.e., new problems are solved by reusing and if necessary adapting the solutions to similar problems that were solved in the past). It has enjoyed considerable success in a wide variety of problem solving tasks and domains. Following a brief overview of the traditional problem-solving cycle in CBR, we examine the cognitive science foundations of CBR and its relationship to analogical reasoning. We then review a representative selection of CBR research in the past few decades on aspects of retrieval, reuse, revision, and retention.
Michael M. Richter and Agnar Aamodt: Case-based reasoning foundations. To appear in: Knowledge Engineering Review, Special Issue on Case-Based Reasoning. 2006. 5 pgs.
A basic observation is that Case-based Reasoning has roots in different disciplines: Cognitive science, knowledge representation and processing, machine learning and mathematics. As a consequence, there are foundational aspects from each of these areas. We briefly discuss them and comment on the relations between these types of foundations.
Cindy Marling, Edwina Rissland and Agnar Aamodt: Integrations with case-based reasoning To appear in: Knowledge Engineering Review, Special Issue on Case-Based Reasoning. 2006. 5 pgs.
This commentary succinctly summarizes work in integrating case-based reasoning (CBR) with other reasoning modalities. Including CBR in mixed mode approaches promotes synergies and benefits beyond those achievable using CBR or other individual reasoning approaches alone. Numerous examples of hybrid systems, with pointers to significant references, are provided.
Frode Srmo, Jrg Cassens, Agnar Aamodt: Explanation in Case-Based Reasoning; Perspectives and Goals . Artificial Intelligence Review. Vol 24, no. 2, October 2005. pp. 109-143.
We present an overview of different theories of explanation from the philosophy and cognitive science communities. Based on these theories, as well as models of explanation from the knowledge-based systems area, a framework for explanation in case-based reasoning (CBR) is presented, based on explanation goals. We propose ways that the goals of the user and system designer should be taken into account when deciding what is a good explanation for a given CBR system. Some general types of goals relevant to many CBR systems are identified, and used to survey existing methods of explanation in CBR. Finally, we identify some future challenges.
Mingyang Gu, Agnar Aamodt: A Knowledge-Intensive Method for Conversational CBR. In: H�ctor Mu�oz Avila, Francesco Ricci (eds.), Case-Based Reasoning Research and Development, 6th International Conference on Casxe-Based Resoning,Proceedings. Chicago, Aug 23-26, 2005. Lecture Notes in Artificial Intelligence 3620, Springer Verlag 2005, pp. 296-311. ISSN 0302-9743.
In conversational case-based reasoning (CCBR), a main problem is how to select the most discriminative questions and display them to users in a natural way to alleviate users� cognitive load. This is referred to as the question selection task. Current question selection methods are knowledge-poor, that is, only statistical metrics are taken into account. In this paper, we identify four computational tasks of a conversation process: feature inferencing, question ranking, consistent question clustering and coherent question sequencing. We show how general domain knowledge is able to improve these processes. A knowledge representation system suitable for capturing both cases and general knowledge has been extended with meta-level relations for controlling a CCBR process. An �explanation-boosted� reasoning approach, designed to accomplish the knowledge-intensive question selection tasks, is presented. An application of our implemented system is illustrated in the car fault detection domain.
Waclaw Kusnierczyk, Agnar Aamodt, Astrid Lgreid: Knowledge-intensive case-based support for automated explanation of biological phenomena. In: 6th International Conference on Case-Based Reasoning, Workshop Proceedings. Chicago, Aug 23-26, 2005. DePaul University, Chicago, 2005. pp. 62-71.
In conversational case-based reasoning (CCBR), a main problem is how to select the most discriminative questions and display them to users in a natural way to alleviate users� cognitive load. This is referred to as the question selection task. Current question selection methods are knowledge-poor, that is, only statistical metrics are taken into account. In this paper, we identify four computational tasks of a conversation process: feature inferencing, question ranking, consistent question clustering and coherent question sequencing. We show how general domain knowledge is able to improve these processes. A knowledge representation system suitable for capturing both cases and general knowledge has been extended with meta-level relations for controlling a CCBR process. An �explanation-boosted� reasoning approach, designed to accomplish the knowledge-intensive question selection tasks, is presented. An application of our implemented system is illustrated in the car fault detection domain.
Agnar Aamodt: Case-based reasoning and intelligent tutoring. Funk, P., Rognvaldsson, T., Xiong, N.(eds.), SAIS-SSLS Proceedings. V�ster�s, Sweden, April 20-22, 2005. M�lardalen H�gskola, V�ster�s, 2005. pp. 8-22. (Invited talk.)
Knowledge-intensive CBR assumes that cases are enriched with general domain knowledge. In the system Creek, developed in our group, there is a strong coupling between cases and general domain knowledge, in that cases are embedded within a semantic network of domain concepts and relations. Although mostly used for decision support, this architecture is also being explored for the purpose of interactive learning support. The theme of the talk is twofold: Integration of cases and general domain knowledge within the knowledge-intensive CBR system Creek, and a discussion of different ways this framework can be used to build intelligent computer-assisted learning systems.
Pl Skalle and Agnar Aamodt: Knowledge-based decision support in oil well drilling. Zhongzhi Shi (ed.), Proceedings of the ICIIP 2004, International Conference on Intelligent Information Systems. Beijing, China, October 21 - 23, 2004.
Oil well drilling is a complex process which frequently is leading to operational problems. The process generates huge amounts of data. In order to deal with the complexity of the problems addressed, and the large number of parameters involved, our approach extends a pure case-based reasoning method with reasoning within a model of general domain knowledge. The general knowledge makes the system less vulnerable for syntactical variations that do not reflect semantically differences, by serving as explanatory support for case retrieval and reuse. A tool, called TrollCreek, has been developed. It is shown how the combined reasoning method enables focused decision support for fault diagnosis and prediction of potential unwanted events in this domain.
Mingyang Gu, Agnar Aamodt, Xin Tong: Component retrieval using conversational case-based reasoning. Zhongzhi Shi (ed.), Proceedings of the ICIIP 2004, International Conference on Intelligent Information Systems. Beijing, China, October 21 - 23, 2004.
Component retrieval, about how to locate and identify appropriate components, is one of the major problems in component reuse. It becomes more critical as more reusable components come from component markets instead of from an in-house component library, and the number of available components is dramatically increasing. In this paper, we review the current component retrieval methods and propose our conversational component retrieval model (CCRM). In CCRM, components are represented as cases, a knowledge-intensive case-based reasoning (CBR) method is adopted to explore context-based semantic similarities between users’ query and stored components, and a conversational case-based reasoning (CCBR) technology is selected to acquire users’ requirements interactively and incrementally.
Agnar Aamodt: Integrating case-specific experiences with general domain knowledge for intelligent information processing. IFIP/ICIIP 2004, International Conference on Intelligent Information Processing, Beijing, October 21-23, 2004. 12 pgs. Invited talk, unpublished, slight modification of the ECCBR 2004 paper.
As single-paradigm reasoning methods become more mature and understood, research activities targeted at combining them become more frequent. Methods that combine case-based reasoning with a model-based component are attracting a growing number of researchers within the case-based reasoning community. There is clear evidence of synergy effects, in that the resulting systems become both more competent and more efficient than if only a single reasoning method is used. The catch is that the burden on the knowledge engineer and domain expert is increased, due to the need for developing an explicit model of general domain knowledge. Methods developed within the knowledge-acquisition and modelling communities, however, as well as work on reusable ontologies, can provide some help. In particular, the notion of knowledge-level modelling has proved to be a useful one in this context. The type of application targeted in our research is decision making in complex environments, such as oil well drilling. Building application systems in such domains further calls for a clarification of basic terms such as “data”, “information”, and “knowledge”, related to their roles in cognitive and computational information processing. For case-based reasoning this becomes particularly relevant, since a case can hold structures of various basic kinds. In the Creek system, developed in our group, there is a strong coupling between cases and general domain knowledge, in that cases are embedded within a general domain model, and the general domain knowledge is used to explain the similarity between two cases that are semantically similar. We are exploring a particular knowledge-level modelling approach for the development of Creek applications. The focusing theme of the talk will be on the utilization of the both case-specific and general domain knowledge, through a combined case-based and model-based reasoning method. This will be made concrete by summarizing the Creek architecture and system, exemplifying its use, and briefly describe a set of recent and ongoing projects where various aspects of this theme are studied.
Agnar Aamodt: Knowledge-intensive case-based reasoning in Creek. Peter Funk, Pedro A. Gonzalez Calero (eds.), Advances in case-based reasoning, 7th European Conference, ECCBR 2004, Proceedings. Madrid, Spain, August/September 2004. Lecture Notes in Artificial Intelligence, LNAI 3155, Spinger, 2004. pgs. 1-15. ISBN0302-9743.
Knowledge-intensive CBR assumes that cases are enriched with general domain knowledge. In CREEK, there is a very strong coupling between cases and general domain knowledge, in that cases are embedded within a general domain model. This increases the knowledge-intensiveness of the cases themselves. A knowledge-intensive CBR method calls for powerful knowledge acquisition and modeling techniques, as well as machine learning methods that take advantage of the general knowledge represented in the system. The focusing theme of the paper is on cases as knowledge within a knowledge-intensive CBR method. This is made concrete by relating it to the CREEK architecture and system, both in general terms, and through a set of example projects where various aspects of this theme have been studied.
Mingyang Gu, Agnar Aamodt: Explanation-boosted question selection in conversational CBR. Pablo Gervas and Kalyan Moy Gupta (eds.), Proceedings of the ECCBR 2004 Workshops, 7th European Conference on Case-Based Reasoning. Madrid, Spain, 30th August - 2nd September 2004. Technocal Report 142-04, Universidad Complutense de Madrid, Departemento de Sistemas Informaticos y Programacion, Madrid, 2004. pgs. 105-114. ISBN 84-89456-36-4
A core research concern in conversational case-based reasoning (CCBR) is how to select the most discriminative and natural questions to askthe user in the conversational process. There are two ways to realize this task: one is to remove the questions whose answers can be inferred from the information a user has provided, which is called dialogue inferencing; the other is to rank the questions to guarantee the most informative questions are asked first, which is referred to as question ranking. In this paper, we present a common explanation-boosted CCBR approach, which utilizes both general domain knowledge and case-specific knowledge to realize dialogue inferencing and question ranking. This approach provides a flexible meta-level knowledge representation model to be able to incorporate richer semantic relations. An application of this approach is illustrated in a car fault detection domain.
Anders Kofod-Petersen, Agnar Aamodt: Case-based situation assessment in a mobile context-aware system. Proceedings of AIMS2003, Workshopon Artificial Intgelligence for Mobil Systems, Seattle,October, 2003.
This paper describes how to utilise Case Based Reasoning for identifying and user situations assessment in a context-aware mobile system. The Case Based Reasoning mechanism attempts to identify what situation the user is in, and utilizes a Multi Agent System, consisting of information suppliers, to provide the user with personalised and context-sensitive information.
Cases are descriptions of situations limited in time and space. The research reported here introduces a method for representation and reasoning with time-dependent situations, or temporal cases, within a knowledge-intensive CBR framework. Most current CBR methods deal with snapshot cases, descriptions of a world state at a single time stamp. In many time dependent situations, value sets at particular time points are less important than the value changes over some interval of time. Our focus is on prediction problems for avoiding faulty situations. Based on a well-established theory of temporal intervals, we have developed a method for representing temporal cases inside the knowledge-intensive CBR system Creek.The paper presents the theoretical foundation of the method, the representation formalism and basic reasoning algorithms, and an example applied to the prediction of unwanted events in oil well drilling.
In this paper we describe work in progress to further develop methods from knowledge-intensive Case-Based Reasoning (CBR) in general, and the Creek system in particular, to address the requirements of Intelligent Tutoring Systems(ITS). ITS systems are often required to solve problems in the domain, but unlike decision-support systems, the ability to communicate the process of solving the problem to the student, is as important as finding the right solution.
Recent work within the CBR community has studied more systematic means of interrelating different types of domain knowledge, tasks, and methods for building and maintaining CBR systems. This paper reviews work from the knowledge acquisition community, targeted at methodologies and tools for analysis, modeling, and maintenance of knowledge components. It is argued that the knowledge level is the appropriate level for describing the behavior of an intended CBR system, and for identifying the contents of its knowledge components. A framework is outlined in which a knowledge-level modeling methodology is adapted for the modeling and maintenance of CBR knowledge contents.
A system that applies a method of knowledge-intensive case-based reasoning, for repairand prevention of unwanted events in the domain of offshore oil well drilling,has been developed in cooperation with an oil company. From several re-occurring problems during oil well drilling the problem of lost circulation, i.e. loss of circulating drilling fluid into the geological formation, was picked out as a pilot problem. An extensive general knowledge model was developed for the domain of oil well drilling. About fifty different cases were created on the basis of information from one North Sea operator. When the completed CBR-system was tested against a new case, five cases with descending similarity were selected by the tool. In an informal evaluation, the two best fitting cases proved to give the operator valuable advise on how to go about solving the new case.
To appropriately utilize the rapidly growing amount of data and information is a big challenge for people and organizations. Standard information retrieval methods, using sequential processing combined with syntax-based indexing and access methods, have not been able to adequately handle this problem. We are currently investigating a different approach, based on a combination of massive parallel processing with case-based (memory-based) reasoning methods. Given the problems of purely syntax-based retrieval methods, we suggest ways of incorporating general domain knowledge into memory-based reasoning. Our approach is related to the properties of the parallel processing microchipMS160, particularly targeted at fast information retrieval from very large datasets. Within this framework different memory-based methods are studied, differing in the type and representation of cases, and in the way that the retrieval methods are supported by explicit general domain knowledge. Cases can be explicitly stored information retrieval episodes, virtually stored abstractions linked to document records, or merely the document records themselves. General domain knowledge can be a multi-relational semantic network, a set of term dependencies and relevancies, or compiled into a global similarity metric. This paper presents the general framework, discusses the core issues involved, and describes three different methods illustrated by examples from the domain of medical diagnosis.
A challenge of future knowledge management and decision support systems is to combine the storage and effective reuse of data, systematically captured as process or system information with user experience in dealing with problems and non-trivial situations In CBR, situation-specific user experiences are typically captured in
cases. In our approach, cases are linked within a semantic network of
more general domain knowledge. In this paper we present a way to
automate the construction and dynamical refinement of such a model of
case-specific and general knowledge, on the basis of external process
data continuously being generated. A data mining method based on a
Bayesian Networks approach is used. We are also looking into how the
notion of causality, being a central issue in both BNs and model-based
combined model.
Frode Srmo, Agnar Aamodt: Knowledge elaboration for improved CBR. In Sixteenth International Joint Conference on Artificial Intelligence, Workshop ML-5: Automating the Construction of Case-Based Reasoners. Stockholm 1999. Sarabjot Singh Anand, Agnar Aamodt, David W. Aha (eds.). pp. 39-43.
When using knowledge-intensive case-based reasoners, a key issue is to keep the knowledge-base updated. This is usually done by having the user update the knowledge during the retain phase, but machine-learning methods can also be used. In this paper we discuss how an elaboration phase, in which plausible inference rules are derived, can be done in a goal-driven manner to increase the system's automatic learning capability
Pinar zturk, Agnar Aamodt: A context model for knowledge-intensive case-based reasoning. International Journal of Human Computer Studies. Vol. 48, 1998. Academic Press. pp 331-355.
Decision-support systems that help solving problems in open and weak theory domains, i.e. hard problems, need improved methods to ground their models in real world situations. Models that attempt to capture domain knowledge in terms of, e.g. rules or deeper relational networks, tend either to become too abstract to be efficient, or too brittle to handle new problems. In our research we study how the incorporation of case-specific, episodic, knowledge enables such systems to become more robust and to adapt to a changing environment by continuously retaining new problem solving cases as they occur during normal system operation. The research reported in this paper describes an extension that incorporates additional knowledge of the problem solving context into the architecture. The components of this context model is described, and related to the roles the components play in an abductive diagnostic process. Background studies are summarized, the context model is explained, and an example shows its integration into an existing knowledge-intensive CBR system.
AgnarAamodt, Helge Langseth: Integrating Bayesian networks into knowledge-intensive CBR. In American Association for Artificial Intelligence, Case-based reasoning integrations; Papers from the AAAI workshop. David Aha, Jody J. Daniels (eds.). Technical Report
In this paper we propose an approach to knowledge intensive CBR, where explanations are generated from a domain model consisting partly of a semantic network and partly of a Bayesian network (BN). The BN enables learning within this domain model based on the observed data. The domain model is used to focus the retrieval and reuse of past cases, as well as the indexing when learning a new case. Essentially, the BN-powered submodel works in parallel with the semantic network model to generate a statistically sound contribution to case indexing, retrieval and explanation.
Pl Skalle, Agnar Aamodt, Jostein Sveen: Case-based reasoning; a method for gaining experience and giving advise on how to avoid and how to free stuck drill strings. Proceedings of IADC Middle East Drilling Conference, Dubai, Nov. 1998.
Stuck pipe seems to be an unavoidable problem. Based on historical data every third well estimated indicates that associated costs exceeds 250 mill $ per year for the industry. The countermeasure is to store the data describing stuck pipe situations and implement procedure similar problems when encountered. When a sufficiently large data base is created, it can be situation: How does the new situation (new case) compare to any of the previous cases and which successful of them? What are the recommendations for the new case? If the new problem is solv supports the recommended solution, or if the solution is different, an additional case can experience. Along with the description of a case-based reasoning system implemented to study presents statistical material of stuck pipe incidents from an operator for the years 1990-19 used to select the parameters for building a hierarchical knowledge model. A realistic case system will help an operator to reduce downtime.
Agnar Aamodt, Helge A. Sandtorv, Ole M. Winnem : CombiningCase Based Reasoning and Data Mining - A way of revealing and reusing RAMS experience.
In Lydersen, Hansen, Sandtorv (eds.),Safety and Reliability; Proceedings of ESREL 98, Trondheim, June 16-19, 1998. Balkena, Rotterdam, 1998. ISBN
90-5410-966-1. pp 1345-1351.
The reuse of previous experience within safety, reliability and maintainability (RAMS) is an important input to complex decision making tasks. Statistical data are vital for many RAMS type of analyses, but frequently the investigation of a more limited number of past cases is preferable, enabling the reuse of relevant and concrete experience in dealing with a new task. In order to effective utilise the increasing amount of information available in computerised databases, a combination of quantitative and qualitative methods are therefore called for. This paper describes an approach where statistical methods for extracting interesting relationships in data (data mining), are combined with knowledge-based methods for capturing and reusing problem solving knowledge in the form of specific experiences (case-based reasoning). The main idea is to combine the two technologies in order to reuse past user experience in dealing with a problem situation, and to extract relevant and focused information from large and scattered data sources. The work is done within the EU project NOEMIE.
Pinar Ozturk, Agnar Aamodt: Towardsa model of context for case-based diagnostic problem solving. In Context-97;Proceedings of the interdisciplinary conference on modeling and using context. Rio de Janeiro, February 1997. pp 198-208.
This paper presents a model of context based on the roles and elements of various context types. Two important roles of context are related to the notions of relevance and focus. The former is important for the quality of the results reached by a problem solving or learning task, while the latter is important for the performance efficiency of the task. Problem solving can be viewed as search in a large problem space where search for different entities is invoked at different stages. Context has a pruning effect on search, increasing proportionally to the incompleteness of the information at hand. Depending on the type of memory structure to be searched for, different types of contexts assist the access to memory. We attempt to give an account of the various types of contexts that facilitate memory access and utilization for different type of tasks. The criteria for distinguishing between several types of context elements are presented, and a context ontology based on these criteria is suggested. We then show how this account is integrated with a case-based approach to clinical problem solving.
Morten Grimnes, Agnar Aamodt: A twolayercase-based reasoning architecture for medical image understanding. In Smith, I., Faltings, B. (eds). Advances in case-based reasoning, (Proc.EWCBR-96), Springer Verlag, Lecture Notes in Artificial Intelligence 1168,1996. pp 164-178.
The paper describes a novel architecture for image understanding. It is based on acquisition of radiologist knowledge, and combines low-level structure analysis with high-level interpretation of image content, within a task-oriented model. A case based reasoner working on a segment case-base contains the individual image segments. These cases with labels are considered indexes for another case based reasoner working on an organ interpretation case base. Both are Creek type case based reasoners, here operating within a propose-critique-modify task structure. Methods for criticizing suggested interpretations by way of explanation, and how interpretations may be modified, are presented. An example run illustrates the system architecture and its key concepts.
Klaus-Dieter Althoff, Agnar Aamodt: Relatingcase-based problem solving and learning methods to task and domain characteristics; towards an analytic framework. AI Communications – The European Journal of Artificial Intelligence, 9 (3) 1966, pp 109-116.
A particular strength of case-based reasoning (CBR) over most other methods is its inherent combination of problem solving with sustained learning through problem solving experience. This is therefore a particularly important topic of study, and an issue that has now become mature enough to be addressed in a more systematic way. To enable such an analysis of problem solving and learning, we have initiated work towards the development of an analytic framework for studying CBR methods. It provides an explicit ontology of basic CBR task types, domain characterisations, and types of problem solving and learning methods. Further, it incorporates within this framework a methodology for combining a knowledge-level, top-down analysis with a bottom-up, case-driven one. In this article, we present the underlying view and the basic approach being taken, the main components of the framework and accompanying methodology, examples of studies recently done and how they relate to the framework.
Klaus-Dieter Althoff, Agnar Aamodt: Zur Analyse fallbasierter Problemlse- und Lernmethode in Abhngigkeit von Charakteristika gegebener Aufgabenstellungen und Anwendungsdomnen. Knstliche Intelligenz (KI), nr. 1 (Mrz) 1966, pp 10-15.(German version of above paper).
Agnar Aamodt: KnowledgeAcquisition and Learning from Experience - The Role of Case-Specific Knowledge,In Gheorge Tecuci and Yves Kodratoff (eds): Machine learning and knowledgeacquisition; Integrated approaches, (Chapter 8),Academic Press, 1995, pp 197-245.As knowledge-based systems are addressing increasingly complex domains, their roles are shifting from classical expert systems to interactive assistants. To develop and maintain such systems, an integration of thorough knowledge acquisition procedures and sustained learning from experience is called for. A knowledge level modeling perspective has shown to be useful for analyzing the various types of knowledge related to a particular domain and set of tasks, and for constructing the models of knowledge contents needed in an intelligent system. To be able to meet the requirements of future systems with respect to robust competence and adaptive learning behavior, particularly in open and weak theory domains, a stronger emphasis should be put on the combined utilization of case-specific and general domain knowledge. In this chapter we present a framework for integrating KA and ML methods within a total knowledge modeling cycle, favoring an iterative rather than a top down approach to system development. Recent advances in the area of case-based reasoning provide a suitable basis. Focus is put on the knowledge maintenance part, for which a case-based method for learning from experience is described and exemplified by existing systems. Our own approach to integration of case-specific and general domain knowledge (the CREEK system) is briefly sketched, and used as a context for discussing case-based approaches to knowledge modeling in general.
Agnar Aamodt, Mads Nygaard: Differentroles and mutual dependencies of data, information, and knowledge - an AI perspective on their integration, Data and Knowledge Engineering 16 (1995), pp 191-222.
The unclear distinction between data, information, and knowledge has impaired their combination and utilization for the development of integrated systems. There is need for a unified definitional model of data, information, and knowledge based on their roles in computational and cognitive information processing. An attempt to clarify these basic notions is made, and a conceptual framework for integration is suggested by focusing on their different roles and frames of reference within a decision-making process. On this basis, ways of integrating the functionalities of databases, information systems and knowledge-based systems are discussed by taking a knowledge level perspective to the analysis and modeling of systems behaviour. Motivated by recent work in the area of case-based reasoning related to decision support systems, it is further shown that a specific problem solving episode, or case, may be viewed as data, information, or knowledge, depending on its role in decision making and learning from experience. An outline of a case-based system architecture is presented, and used to show that a focus on the retaining and reuse of past cases facilitates a gradual and evolutionary transition from an information system to a knowledge-based system.
Agnar Aamodt: Explanation-driven case-based reasoning, In S. Wess, K. Althoff, M. Richter (eds.): Topicsin Case-based reasoning. Springer Verlag, 1994. pp274-288.
Problem solving in weak theory domains should compensate for the lack of strong theories by combining the various other knowledge types involved. Such methods should be able to effectively combine general domain knowledge with specific case knowledge. A method is described that utilises a presumably extensive and dense model of general domain knowledge as explanatory support for case-based problem solving and learning. A generic reasoning method - captured in what is called the activate-explain-focus cycle - is able to utilise a rich knowledge model in producing context-dependent explanations. A specialisation of this method for each of the main subprocesses of case-based reasoning is presented, and illustrated with examples.
Agnar Aamodt: AKnowledge Representation System for Integration of General and Case-SpecificKnowledge. Proceedings from IEEE TAI-94, International Conference onTools with Artificial Intelligence. New Orleans,November 5-12, 1994. 4 pages.
Combining various knowledge types - and reasoning methods - in knowledge-based systems is a challenge to the knowledge representation task. The paper describes an object-oriented, frame-based knowledge representation system aimed at unifying case-specific and general domain knowledge within a single representation system. It is targeted at the representational needs that have emerged from research in knowledge-intensive case-based reasoning, addressing complex problem solving in open and weak theory domains. Emphasis is put on representational expressiveness, on flexible reasoning and control schemes, and on easy inspection of cases and other knowledge objects.
Agnar Aamodt, Enric Plaza: Case-based reasoning; Foundational issues, methodological variations, and system approaches (ftp-version, html-version)AI Communications, Vol.7, No.1, March 1994, pp. 39-59.
Case-based reasoning is a recent approach to problem solving and learning that has got a lot of attention over the last few years. Originating in the US, the basic idea and underlying theories have spread to other continents, and we are now within a period of highly active research in case-based reasoning in Europe, as well. This paper gives an overview of the foundational issues related to case- based reasoning, describes some of the leading methodo- logical approaches within the field, and exemplifies the current state through pointers to some systems. Initially, a general framework is defined, to which the subsequent descriptions and discussions will refer. The framework is influenced by recent methodologies for knowledge level descriptions of intelligent systems. The methods for case retrieval, reuse, solution testing, and learning are summa-rized, and their actual realization is discussed in the light of a few example systems that represent different CBR approaches. We also discuss the role of case-based methods as one type of reasoning and learning method within an integrated system architecture.
Mads Nygård, Agnar Aamodt: Road transport informatics; Conceptual framework and technological components. In: VNIS '93, 4thInternational Conference on Vehicle Navigation and Information Systems. Ottawa,Canada, October 12-15 1993.
This paper addresses an advanced utilization of information technology within the road transport field. The main motivation for our investigation is the synergetic effect a genuine coupling of the two fields will create. A major objective is to build a conceptual framework for road transport informatics in order to establish a common basis for researchers from both fields. Another major objective is to identify and specify suitable mechanisms for implementation of key system components. The main paradigm for our work is always how mechanisms from the information technology field can optimize support needs within road transport.
Agnar Aamodt: Acase-based answer to some problems of knowledge-based systems. In E.Sandewal and C.G. Jansson (eds): Scandinavian Conference on ArtificialIntelligence 1993. IOS Press, 1993. pp. 168-182.
Among the most important challenges for contemporary AI research are the development of methods for improved robustness, adaptability, and overall interactiveness of systems. Interactiveness, the ability to perform and react in tight co-operation with a user and/or other parts of the environment, can be viewed as subsuming the other two. There are various approaches to addressing these problems, spanning from minor improvements of existing methods and theories, through new and different methodologies, up to completely different paradigms. As an example of the latter, the very foundation of knowledge-based systems, based on a designer's explicit representation of real world knowledge in computer software structures, has recently been questioned by prominent members of the KBS community. In the present paper, some foundational issues of the knowledge-based paradigm are reviewed, and the main arguments of the critiquing position are discussed. Some of the deficiencies of current approaches pointed out by the critics are acknowledged, but it is argued that the deficiencies cannot be solved by escaping from the knowledge-based paradigm. However, an alternative to the main-stream, generalisation-based, approach is needed. An approach is advocated where reasoning from situation-specific knowledge - captured as a collection of previously solved cases, is combined with generalised domain knowledge in the form of a densely connected semantic network. The approach enables continuos, sustained learning by updating the case base after each new problem has been solved. The paper shows how an example of this approach - the CREEK system - can provide an answer within the knowledge-based paradigm to the problems pointed out by the critics.
Enric Plaza, Agnar Aamodt, Ashwin Ram, Walter van de Velde, Maarten van Someren: Integrated learning architectures. In: Machine Learning: ECML-93, European Conference on Machine Learning, Vienna, Austria, April5-7 1993. Springer Verlag, 1993. pp.429-441.
Research in systems where learning is integrated to other components like problem solving, vision, or natural language is becoming an important topic for Machine Learning. Situations where learning methods are embedded or integrated into broader systems offers new theoretical challenges to ML and enlarge the potential range of ML applications. In this position paper we propose the research topic of integrated learning architectures as an initial discussion of the role of learning in intelligent systems. We review the current state of the art and characterise several dimensions along which integrated learning architectures may vary. This paper has been prepared as a position paper with the purpose of providing an initial common ground for discussion in the ECML-93 Workshop on Integrated Learning Architectures. The paper has been edited by E Plaza on the basis of the individual contributions of the authors.
Ingeborg Solvberg, Inge Nordbø, Agnar Aamodt: Knowledge-based information retrieval. Future Generation Computer Systems, 7 (1991/92) pp379-390.
A knowledge-based system is used as a front-end to a very large database to increase the relevance of the information being retrieved. The subject domain of the database is modelled in a semantic network and the queries to the database are expanded according to the semantic model. An experiment has been performed on a bibliographic database, by developing the prototype KNOWIT, a knowledge-based front-end to the information retrieval system ESA-QUEST. An experimental evaluation shows that the number of relevant bibliographic references retrieved with the knowledge-based front-end is significantly improved, without compromising the precision of the retrieval.
Geir Aakvik, Agnar Aamodt, Inge Nordbø: A knowledge Representation FrameworkSupporting Knowledge Modelling. EKAW- 91, Fifth European Knowledge Acquisition for Knowledge-based Systems Workshop, Crieff,Scotland, June 1991.
The paper points out the advantages of using a flexible, expressive and extendible frame representation system in domain knowledge modeling, and how this system also can be used to design and implement a runable system. In this way, the 'knowledge transformation gap' between an intermediate knowledge model and an operational one is shrunk. Various types of knowledge are extracted during the knowledge elicitation activity and then structured into an intermediate knowledge model, which forms the basis for system design and implementation. The proposed common representation framework is a frame-based system, emphasizing a thorough explicit modelling of various types of knowledge. A short distance between the internal representation and a user readable format facilitates the use of graphical editors both for knowledge modeling and editing of the target system knowledge. Further, this also facilitates knowledge maintenance methods incorporating semi-automatic learning from experience.
Agnar Aamodt: Knowledge-intensive case-based reasoning and learning. ECAI-90,Ninth European Conference on Artificial Intelligence, Stockholm, August 1990.
In case-based reasoning (cbr) a problem is solved by matching the problem description to a previously solved case, using the past solution in solving the new problem. A case-based reasoner learns after each problem solving session by retaining relevant information from a problem just solved, making the new experience available for future problem solving. Crucial steps in a cbr process include finding a good match to a new problem, adapting a previous solution to successfully solve the new problem, and deciding how to index and store a new case for later effective retrieval. Previous cbr systems relied on syntactic rather than semantic or pragmatic criteria in performing these steps. A comprehensive model of general domain knowledge is needed in order to match cases based on their meaning contents. This has lead to systems that attempt to combined case-based methods with model-based - explanation-based - approaches. Two systems representative of this research, PROTOS and CASEY, are briefly described. The systems are discussed with respect to what type of general knowledge they contain, and the degree of model-based support for the case-based reasoning and learning processes they exhibit. Some of their weaknesses are identified, and an improved approach to integration of case-based and model-based methods - called CREEK - is suggested.
Agnar Aamodt: A computational model of knowledge-intensive problem solving and learning. In: Bob Wielinga, et al. (eds.): Current trends in knowledge acquisition. IOS Press, Amsterdam, June 1990.
If knowledge-based systems are to become more competent and robust in solving real world problems, they need to be able to adapt to an evolving domain and a changing environment. This paper proposes a computational model - a framework -for knowledge-intensive problem solving and learning from experience. The model has been instantiated in an architecture for knowledge-intensive case-based reasoning and learning called creek (case-based reasoning through extensive expert knowledge). The importance of a thorough, extensive knowledge model to support the reasoning and learning processes is emphasized. In case-based reasoning a problem is solved by retrieving a similar past problem case, and using this case in solving the new problem. Learning becomes a process of extracting relevant information from a problem just solved, and integrating the new case into the existing case-base. The computational model presented combines case-based learning and reasoning with model-based and rule-based methods. The type of problems addressed are problems within open, weak theory domains.
Agnar Aamodt: Robust expert systems that learn from experience - An architectural framework. EKAW-89, Third European Knowledge Acquisition for Knowledge-based Systems Workshop, Paris, July 1989. pp.311-326.
This work focuses on two major weaknesses of current knowledge-based systems: Their brittle competence and their inability to learn from experience. As a step towards attacking these problems, a KBS architecture that integrates case-based reasoning and learning into a more general knowledge acquisition and problem solving framework is proposed. The importance of developing a conceptual knowledge model of the domain - tightly interconnecting concept-descriptions, relations, constraints, etc. is emphasized. This conceptual model should look into bordering domains of the actual application as well as parts of the general world related to the domain. We are developing a system - called CREEK - to support the building and continuous updating of diagnosis type expert systems along these lines. The learning method involved is knowledge intensive, case-based learning. Learning becomes a part of the problem solving process, thereby obtaining incremental learning as a natural part of the system's problem solving effort.
Agnar Aamodt, Jorun Eggen, A. MacDonald, S. Johnsen: Modelling sedimentological expertise: the FACEXP project. In: D. Sririam, R.A. Adey (eds.): Knowledge Based Systems for engineering, classification and control. Computational Mechanics Publications, 1987. pp. 397-413.
A general framework for modeling of expertise is presented. Expertise is knowledge and the way knowledge is used to solve expert problems. The framework forms the basis for a conceptual expertise model of sedimentary environment interpretation. The FACEXP system is an expert system to assist a geologist in interpreting sedimentary environments from borehole sequences. Two prototypes have been developed - one using OPS83 on a VAX, and the other using the KEE/Zetalisp environment on an Explorer Lisp Machine.
Last modified: April 2006