This paper describes a successful approach to concept learning for heuristic classification. Almost all current programs for this task create or use explicit, abstract generalizations. These programs are largely ineffective for domains with weak or intractable theories. An exemplar-based approach is suitable for domains with inadequate theories but raises two additional problems: determining similarity and indexing exemplars. Our approach extends the exemplar-based approach with solutions to these problems. An implementation of our approach, called Protos, has been applied to the domain of clinical audiology. After reasonable training, Protos achieved a competence level equaling that of human experts and far surpassing that of other machine learning programs. Additionally, an "ablation study" has identified the aspects of Protos that are primarily responsible for its success. 1 Introduction This paper describes a successful approach to the task of concept learning for heuristic clas...

The use of predefined models of problem-solving methods is receiving considerable attention from researchers in the area of knowledge acquisition. Using these models, developers of knowledge-acquisition tools are able to prescribe the roles in which knowledge is used in completing a given task. A number of method-oriented architectures based on a single problem-solving method have been developed by various research groups. Because the methods are domain-independent, method-oriented architectures are limited by the fact that knowledge roles that depend on domain-specific considerations cannot be represented using the model of problem solving. In addition, the interface between the knowledge-acquisition tool and the application expert cannot adequately convey the role of each knowledge type in the task model. PROTG-II is a knowledge-acquisition shell that we are building to generate knowledge-acquisition tools automatically without presupposing a specific model of problem-solving. The shell manages a library of mechanisms---procedures of grain size smaller than that of problem-solving methods. Mechanisms can be combined in PROTG-II to construct problem-solving methods and to define the roles of knowledge that depend on domain considerations. Furthermore, PROTG-II utilizes the concept of adaptation in interfaces to allow the knowledge engineer to produce interfaces that are task- and domain-specific. In this paper, we present the PROTG-II shell and examine the components of its architecture. We also demonstrate the use of PROTG-II with a running example, and discuss the design techniques used to overcome the limitations of method-specific architectures.

PROTG-II is a suite of tools and a methodology for building knowledge-based systems and domain-specific knowledge-acquisition tools. In this paper, we show how PROTG-II can be applied to the task of providing protocol-based decision support in the domain of treating HIVinfected patients. For this task, we use a problem-solving method called episodic skeletal-plan refinement. This method is decomposable; we construct it from a set of reusable components. In addition, we build an application ontology that consists of the terms and relations in the domain, plus terms that supply method-specific knowledge requirements. From this ontology, we automatically generate a domain-specific knowledge-acquisition tool. The general goal of the PROTG-II approach is to produce systems and components that are easily maintained and reusable. This is the rationale for constructing a problem-solving method from a set of smaller-grained methods and mechanisms. This is also why our knowledge-acquisition tools are domain-specific and generated automatically from ontologies. Although our evaluation is still preliminary, for the application task of providing protocol-based decision support, we show that these goals of reusability and easy maintenance can be achieved. We discuss design decisions and the tradeoffs that have to be made in the development of the system. Keywords. Decision support; expert systems; knowledge acquisition.

Metalevel tools can support the software development process by automating the design of task- and application-specific tools. Dash is a metalevel tool that allows developers to generate domain-specific knowledge-acquisition tools from domain ontologies. Domain specialists use the knowledge-acquisition tools generated by dash to instantiate the concepts and relationships defined in the domain ontologies. The output of the knowledge-acquisition tools is a collection of instances that constitute the knowledge base for a knowledge-based system.

Building a problem solver and acquiring the knowledge needed to operate it are the two central goals of knowledge engineering. To achieve these goals, knowledge engineers construct models of the domain and of the task of interest. The various approaches used for modeling, however, have so far failed to define methods and techniques that can be applied across domains and tasks, and to produce models that can be reused in future applications. In this paper, we propose that both of these objectives can be achieved by the use of building blocks called mechanisms. We examine the composition of mechanisms and also show how these mechanisms can be manipulated to construct problem-solving methods. We present PROTG-II, a knowledge-acquisition shell that uses problem-solving methods to drive the modeling of tasks, the automatic generation of knowledge-acquisition tools, and the control flow of the problem solver. The modeling of tasks, within the context of PROTG-II, is illustrated with two examples: one from the game domain and another from the medical-therapy domain. In addition, we introduce the conceptual basis for a library of mechanisms that serves as a repository of reusable knowledge components.

A system that acquires knowledge from a user should be able to reflect upon the knowledge that it has---at each moment---and understand what kinds of new knowledge it needs to learn. For the past two decades, research in the area of knowledge acquisition has been moving towards systems that have access to richer representations of knowledge about their task. This paper reviews some well-known knowledge acquisition tools representative of this trend. It also describes our recent work in EXPECT, a system with explicit representations of knowledge about the task and the domain that supports knowledge acquisition for a wider range of tasks and applications than its predecessors. We hope our observations will be useful to researchers in user interfaces and in machine learning concerned with acquiring information from users. Introduction The acquisition of knowledge about a task can be viewed as a process of incorporating new knowledge into some existing knowledge structure [Rosenbl...

Method-oriented knowledge-acquisition tools are based on a model, or method, of problem solving and can acquire knowledge for the class of tasks that can be solved with that particular problem-solving method. The capture of knowledge takes place in knowledge editors. These editors are typically based on the individual tool's domainindependent method; they fail to reflect task- and domainspecific characteristics and have no ability to adapt to user requirements. Mecano is a user-interface management system that generates automatically adaptable knowledge editors for the PROTG-II knowledge-acquisition shell. Mecano allows knowledge engineers to specify the components of a knowledge editor independently of any underlying problem-solving method. It also provides facilities for constraining the operations allowed on the components, for selecting interaction styles for each component, and for linking components to coordinate their simultaneous display. Knowledge editors generated by Mecano take into account the needs and requirements of given tasks, domains, and users, and guide the users through the knowledge-editing process by providing visual cues and by limiting the permissible editing operations to those relevant in the domain of interest.

An interviewer has two kinds of knowledge. One is about a domain under consideration and the other is knowledge for interview itself which makes the interviewer an expert of interview. The latter seems to be independent of the domain because an experienced interviewer, such as a TV interviewer, can carry on his tasks in many fields. Furthermore, we believe that the interview knowledge consists of several interview knowledge primitives. Based on this idea, we are developing SIS, a Shell for Interview Systems which has seven question strategy primitives as the knowledge for interviewing. Generality and effectiveness of SIS are shown through two implementation examples of interview systems, I 2 s and MORE. The seven primitives are shown to be efficient for the two system whose domains and tasks are entirely different. 1

All important features, such as usability, quality, or efficiency of a product, are determined early on in the design process by the extent to which design requirements are captured and understood. Traditionally, however, those most effected by design decisions, such as users, have been left out of the design process in general and the requirement gathering in particular.