Showing papers by "Qiang Yang published in 1999"

Remembering to add: competence-preserving case-addition policies for case-base maintenance

Abstract: Case-base maintenance is gaining increasing recognition in research and the practical applications of case-based reasoning (CBR). This intense interest is highlighted by Smyth and Keane's research on case deletion policies. In their work, Smyth and Keane advocated a case deletion policy, whereby the cases in a case base are classified and deleted based on their coverage potential and adaptation power. The algorithm was empirically shown to improve the competence of a CBR system and outperform a number of previous deletion-based strategies. In this paper, we present a different case-base maintenance policy that is based on case addition rather than deletion. The advantage of our algorithm is that we can place a lower bound on the competence of the resulting case base; we demonstrate that the coverage of the computed case base cannot be worse than the optimal case base in coverage by a fixed lower bound, and the coverage is often much closer to optimum. We also show that the Smyth and Keane's deletion based policy cannot guarantee any such lower bound. Our result highlights the importance of finding the right case-base maintenance algorithm in order to guarantee the best case-base coverage. We demonstrate the effectiveness of our algorithm through an experiment in case-based planning.

Activating CBR Systems Through Autonomous Information Gathering

Abstract: 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

Dynamic Refinement of Feature Weights Using Quantitative Introspective Learning

Abstract: Recently more and more researchers have been supporting the view that learning is a goaldriven process. One of the key properties of a goal-driven learner is introspectiveness-the ability to notice the gaps in its knowledge and to reason about the information required to fill in those gaps. In this paper, we introduce a quantitative introspective learning paradigm into case-based reasoning (CBR). The result is an integrated problem-solving model which will learn introspectively feature weights in a case base in order to be responsive dynamically to its users. In contrast to the existing qualitative methods for introspective learning, our model has the advantage of being able to capture accurate learning information in the interactions with its users. A CBR system equipped with quantitative introspective learning ability can allow the feature weights to be captured automatically and to track its users' changing preferences continuously. In such a system, while the reasoning part is still case-based, the learning part is shouldered by a quantitative introspective learning model. Weight learning and evolution are accomplished in the background. The effectiveness of this integration will be demonstrated through a series of empirical experiments.

A catalog of agent coordination patterns

Abstract: This paper surveys the current state of the art in agentoriented software engineering, focusing on the area of coordinated multi-agent systems. In multi-agent systems, the interactions between the agents are crucial in determining the effectiveness of the system. Hence the adoption of an appropriate coordination mechanism is pivotal in the design of multi-agent system architectures. This paper does not focus on agent theory, rather on the development of an agent-oriented software engineering methodology, collaboration architectures and design patterns for collaboration. A catalog of coordination patterns inherent in multi-agent architectures is presented. Such patterns may be utilized in the architectural design stage of an agent-oriented software engineering methodology.

Plan Mining by Divide-and-Conquer

Abstract: Plans or sequences of actions are an important form of data With the proliferation of database technology plan databases or planbases are increasingly common E cient discovery of important patterns of actions in plan databases presents a challenge to data mining In this paper we present a method for mining signi cant patterns of successful actions in a large planbase using a divide and conquer strategy The method exploits multi dimensional generalization of sequences of actions and extracts the inherent hierarchical structure and sequential patterns of plans at di erent levels of abstraction These patterns are used in turn to subsequently narrow down the search for more speci c patterns The process is analogous to the use of divide and conquer methods in hierarchical planning We illustrate our approach using a travel planning database

Is CBR Applicable to the Coordination of Search and Rescue Operations? A Feasibility Study

Abstract: 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.

Real-Time Scheduling for Multi-agent Call Center Automation

Abstract: In a call center, service agents with different capabilities are available for solving incoming customer problems at any time. To supply quick response and better problem solution to customers, it is necessary to schedule customer problems to appropriate service agents efficiently. We developed SANet, a service agent network for call center, which integrates multiple service agents including both software agents and human agents, and employs a broker to schedule customer problems to service agents for better solutions according to their changing capabilities and availability. This paper describes the real-time scheduling method in SANet as well as its architecture. There are two phases in our scheduling method. One is problem-type learning. The broker is trained to learn the problem types and hence can decide the type of incoming problems automatically. The other is the scheduling algorithm based on problem types, capabilities and availability of service agents. We highlight an application in which we apply SANet to a call center problem for a cable-TV company. Finally, we support our claims via experimental results and discuss related works.

Maintaining Large Case Bases Using Index Learning and Clustering

Abstract: In a typical case based reasoning application, the case bases grow at a very fast rate and their contents become increasingly diverse, making it necessary to partition a large case base into several smaller ones. Their users are overloaded with vast amounts of information during the retrieval process. These problems call for the development of eeective case-base maintenance methods. In this paper we present two major case-base maintenance methods. The rst method partitions cases into clusters where the cases in the same cluster are more similar than cases in other clusters. In addition to the content of textual cases, the clustering method we propose can also be based on values of attributes that may be attached to the cases. Clusters can be converted to new case bases, which are smaller in size and when stored distributedly, can entail simpler maintenance operations. The contents of the new case bases are more focused and easier to retrieve and update. Our second method allows case-base indexes to be continuously learned and updated. We use a revised version of neural network learning algorithm to maintain the case base indexes which demonstrates exceptional error-convergence property.