Ray Bareiss - Academia.edu (original) (raw)
Papers by Ray Bareiss
Routledge eBooks, May 23, 2019
ACM Conference on Hypertext, 1996
National Conference on Artificial Intelligence, Jul 11, 1993
At the Institute for the Learning Sciences we have been developing large scale hypermedia systems... more At the Institute for the Learning Sciences we have been developing large scale hypermedia systems, called ASK systems, that are designed to simulate aspects of conversations with experts. They provide access to manually indexed, multimedia databases of story units. We are particularly concerned with finding a practical solution to the problem of finding indices for thes units when the database grows too large for manual techniques. Our solution is to provide automated assistance that proposes relative links between units, eliminating the need for manual unit-to-unit comparison. In this paper we describe eight classes of links, and show a representation and inference procedure to assist in locating instances of each.
Elsevier eBooks, 1993
Problem CategoriesThe abstract problem categories include a set of concepts takenfrom Andersen Co... more Problem CategoriesThe abstract problem categories include a set of concepts takenfrom Andersen Consulting's change management methodology andanother set of common sense categories borrowed from the study ofconventional, proverbial, wisdom. These categories provide anexplicit way of organizing both the sets of problem-describingfeatures, and the cases which are indexed by those features.The abstract problem categories drawn from the changemanagement methodology are called
Artificial Intelligence, Sep 1, 1990
This paper describes a successful approach to concept learning for heuristic classi cation. Almos... more This paper describes a successful approach to concept learning for heuristic classi cation. Almost all current programs for this task create or use explicit, abstract generalizations. These programs are largely ine ective 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 identi ed the aspects of Protos that are primarily responsible for its success.
Machine Learning, 1991
Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquis... more Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquisition system. The Protos project did result in the development and distribution of such a system. However, the motivation for the project was to formulate and test a unified approach to representing, acquiring, and reasoning with polymorphic concepts. I believe this aspect of the research will be of more enduring interest to the Machine Learning community than the Protos knowledge acquisition system itself. Reich correctly points out the difficulties of evaluating a complex system and understanding its behavior. Our experiments, summarized in Porter et al. (1990), involved five people over the course of two years. Even so, these experiments left many important questions unanswered. I can, however, answer some of the specific questions asked in Section 3.1:
![Research paper thumbnail of Every curriculum tells a story [e-commerce master's program]](https://attachments.academia-assets.com/107370412/thumbnails/1.jpg)
](
Elsevier eBooks, 1989
Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquis... more Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquisition system. The Protos project did result in the development and distribution of such a system. However, the motivation for the project was to formulate and test a unified approach to representing, acquiring, and reasoning with polymorphic concepts. I believe this aspect of the research will be of more enduring interest to the Machine Learning community than the Protos knowledge acquisition system itself. Reich correctly points out the difficulties of evaluating a complex system and understanding its behavior. Our experiments, summarized in Porter et al. (1990), involved five people over the course of two years. Even so, these experiments left many important questions unanswered. I can, however, answer some of the specific questions asked in Section 3.1:
The Journal of the Learning Sciences, 1992
Telling stories is an effective way to teach aspects of nearly every task and domain. However, to... more Telling stories is an effective way to teach aspects of nearly every task and domain. However, to be effectively remembered, a story must be told in a context that enables the hearer to index it functionally in memory. This occurs naturally when stories are told to students ...
Human Factors in Computing Systems, 1993
Springer eBooks, 2013
This paper describes the design and initial evaluation of a mobile application for training Commu... more This paper describes the design and initial evaluation of a mobile application for training Community Emergency Response Teams. Our goal is to model the kind of remediation and performance support provided in high-end eLearning systems, and provide it during hands-on learning in the real world, using mobile phones and sensors embedded in the environment. Thus far we have designed the learning system and tested it with real users, simulating sensor-based activity recognition using an Android-based Wizard of Oz system that we have developed. Our initial user tests found that users were able to use the system to complete tasks, including some that they had never done before. They had little difficulty understanding the interaction mechanism, and overall reacted positively to the system. Though learner reaction was generally positive, these user tests yielded important feedback about ways we can better manage the division between the real world and the digital world.
... which captures the experience of officers responsible for planning and coordinating military ... more ... which captures the experience of officers responsible for planning and coordinating military transportation during Desert Shield and Desert Storm (D/S/S ... We also thank Roger Schank andBill Ferguson for their valuable insights into how to structure the Interview Coach project. ...
exemplar based knowledge acquisition a unified approach to the dynamics of law browserfame all th... more exemplar based knowledge acquisition a unified approach to the dynamics of law browserfame all the sweet promises fakof rock cycle test study guide pfd answers niraj elseviers dictionary of nuclear engineering jamroz la garantia en el estado constitucional de derecho ppap guide ebook | dantua israel book of the 20th century fakyu marine corps embassy security group training and readiness springboard 10th grade sesog a quiet crisis in america auzww the eucharist 50 questions from the pews ebook board leadership no 81 september october 2005 study guide for essentials of maternity newborn and womens urban ministry in a new millenium benjay red moon blood moon eclipses full color version ebook golf and philosophy lessons from the links philosophy of cummins 4bt rebuild manual tgdo from parador to parador spain tourist paradores summers end the clan maclean ballad romances first aid for depression 5320b sesog international studies in taxation law and economics selbstmanagement ressourcenorientiert maja storch ebook inncom e528 thermostat avexfx tennessee medicaid manual 2015 wlets the evolution of consciousnessas revealed through the patient has the floor zaraa exemplar for tourism 2014 helano jcb compact tractor service manual abilantis komatsu pc200 6 factory service repair manual ebook | www chemistry essay answers trupin saving energy growing jobs how environmental protection history of the war in hungary in 1848 and 1849 benjay 1935 1942 packard 120 body repair shop manual reprint bouquet of thorns valcap manual nissan quest 1997 billballam cummins 903 service manual coreysmith ford falcon xc workshop manual towies ergonomics foundational principles applications and air sparging for site remediation ebook | dr-calorie
At the Institute for the Learning Sciences we have been developing large scale hypermedia systems... more At the Institute for the Learning Sciences we have been developing large scale hypermedia systems, called ASK systems, that are designed to simulate aspects of conversations with experts. They provide access to manually indexed, multimedia databases of story units. We are particularly concerned with finding a practical solution to the problem of finding indices for thes units when the database grows too large for manual techniques. Our solution is to provide automated assistance that proposes relative links between units, eliminating the need for manual unit-to-unit comparison. In this paper we describe eight classes of links, and show a representation and inference procedure to assist in locating instances of each.
Our research concerns the construction of knowledge-rich memories, in hypermedia form, for use as... more Our research concerns the construction of knowledge-rich memories, in hypermedia form, for use as aids to problem-solving. One of the most difficult steps in building such memories is constructing a rich set of links between the content elements they contain (i.e., text segments, graphics, and video clips). This paper describes point linking, a method we have developed for automated linking. Point linking is currently being incorporated into the ASKTool, a hypermedia editor in ongoing use at the Institute for the Learning Sciences.
Elsevier eBooks, 1989
This chapter provides an overview of how Protos represents concepts, performs classification, and... more This chapter provides an overview of how Protos represents concepts, performs classification, and learns. Protos represents concepts extensionally as collections of exemplars. Accompanying domain knowledge provides cohesiveness within these categories. Classification is a two-step process. The features of a new case provide hypotheses as to its classification. Protos seeks to verify a hypothesis by retrieving an appropriate exemplar and constructing an explanation of its similarity to the new case. When it cannot classify a new case or cannot explain a classification, Protos learns by interacting with its teacher. It selectively retains exemplars and explanations and formulates indices, which allow efficient retrieval during future problem solving. The Protos approach attempts to expand the bandwidth of communication between the teacher and learning system to reduce the amount of training required for concept acquisition. Protos autonomously learns the things that it can reasonably be expected to learn. It depends upon the interaction with the teacher to acquire the knowledge that is outside of the scope of its abilities.
Elsevier eBooks, 1989
This chapter presents the design of Protos, explaining how the Protos approach has been instantia... more This chapter presents the design of Protos, explaining how the Protos approach has been instantiated as a concept representation and a set of algorithms. The chapter elaborates and follows the organization of Chapter 2. First, the representation of Protos' category structure is explained. The featural representation of cases is described, and Protos' explanation language is defined. Second, the classification procedure is explained. Hypothesis formation, exemplar selection, knowledge-based pattern matching, and similarity assessment are described. Third, the details of Protos' learning procedures are presented. The selective retention of exemplars, integration of explanations into domain knowledge, and creation and revision of indices are described. Detailed examples of the classification and learning procedures in action are presented in Chapter 4.
Routledge eBooks, May 23, 2019
ACM Conference on Hypertext, 1996
National Conference on Artificial Intelligence, Jul 11, 1993
At the Institute for the Learning Sciences we have been developing large scale hypermedia systems... more At the Institute for the Learning Sciences we have been developing large scale hypermedia systems, called ASK systems, that are designed to simulate aspects of conversations with experts. They provide access to manually indexed, multimedia databases of story units. We are particularly concerned with finding a practical solution to the problem of finding indices for thes units when the database grows too large for manual techniques. Our solution is to provide automated assistance that proposes relative links between units, eliminating the need for manual unit-to-unit comparison. In this paper we describe eight classes of links, and show a representation and inference procedure to assist in locating instances of each.
Elsevier eBooks, 1993
Problem CategoriesThe abstract problem categories include a set of concepts takenfrom Andersen Co... more Problem CategoriesThe abstract problem categories include a set of concepts takenfrom Andersen Consulting's change management methodology andanother set of common sense categories borrowed from the study ofconventional, proverbial, wisdom. These categories provide anexplicit way of organizing both the sets of problem-describingfeatures, and the cases which are indexed by those features.The abstract problem categories drawn from the changemanagement methodology are called
Artificial Intelligence, Sep 1, 1990
This paper describes a successful approach to concept learning for heuristic classi cation. Almos... more This paper describes a successful approach to concept learning for heuristic classi cation. Almost all current programs for this task create or use explicit, abstract generalizations. These programs are largely ine ective 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 identi ed the aspects of Protos that are primarily responsible for its success.
Machine Learning, 1991
Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquis... more Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquisition system. The Protos project did result in the development and distribution of such a system. However, the motivation for the project was to formulate and test a unified approach to representing, acquiring, and reasoning with polymorphic concepts. I believe this aspect of the research will be of more enduring interest to the Machine Learning community than the Protos knowledge acquisition system itself. Reich correctly points out the difficulties of evaluating a complex system and understanding its behavior. Our experiments, summarized in Porter et al. (1990), involved five people over the course of two years. Even so, these experiments left many important questions unanswered. I can, however, answer some of the specific questions asked in Section 3.1:
Elsevier eBooks, 1989
Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquis... more Yorarn Reich has chosen to discuss Protos (Bareiss, 1989) solely as an automated knowledge acquisition system. The Protos project did result in the development and distribution of such a system. However, the motivation for the project was to formulate and test a unified approach to representing, acquiring, and reasoning with polymorphic concepts. I believe this aspect of the research will be of more enduring interest to the Machine Learning community than the Protos knowledge acquisition system itself. Reich correctly points out the difficulties of evaluating a complex system and understanding its behavior. Our experiments, summarized in Porter et al. (1990), involved five people over the course of two years. Even so, these experiments left many important questions unanswered. I can, however, answer some of the specific questions asked in Section 3.1:
The Journal of the Learning Sciences, 1992
Telling stories is an effective way to teach aspects of nearly every task and domain. However, to... more Telling stories is an effective way to teach aspects of nearly every task and domain. However, to be effectively remembered, a story must be told in a context that enables the hearer to index it functionally in memory. This occurs naturally when stories are told to students ...
Human Factors in Computing Systems, 1993
Springer eBooks, 2013
This paper describes the design and initial evaluation of a mobile application for training Commu... more This paper describes the design and initial evaluation of a mobile application for training Community Emergency Response Teams. Our goal is to model the kind of remediation and performance support provided in high-end eLearning systems, and provide it during hands-on learning in the real world, using mobile phones and sensors embedded in the environment. Thus far we have designed the learning system and tested it with real users, simulating sensor-based activity recognition using an Android-based Wizard of Oz system that we have developed. Our initial user tests found that users were able to use the system to complete tasks, including some that they had never done before. They had little difficulty understanding the interaction mechanism, and overall reacted positively to the system. Though learner reaction was generally positive, these user tests yielded important feedback about ways we can better manage the division between the real world and the digital world.
... which captures the experience of officers responsible for planning and coordinating military ... more ... which captures the experience of officers responsible for planning and coordinating military transportation during Desert Shield and Desert Storm (D/S/S ... We also thank Roger Schank andBill Ferguson for their valuable insights into how to structure the Interview Coach project. ...
exemplar based knowledge acquisition a unified approach to the dynamics of law browserfame all th... more exemplar based knowledge acquisition a unified approach to the dynamics of law browserfame all the sweet promises fakof rock cycle test study guide pfd answers niraj elseviers dictionary of nuclear engineering jamroz la garantia en el estado constitucional de derecho ppap guide ebook | dantua israel book of the 20th century fakyu marine corps embassy security group training and readiness springboard 10th grade sesog a quiet crisis in america auzww the eucharist 50 questions from the pews ebook board leadership no 81 september october 2005 study guide for essentials of maternity newborn and womens urban ministry in a new millenium benjay red moon blood moon eclipses full color version ebook golf and philosophy lessons from the links philosophy of cummins 4bt rebuild manual tgdo from parador to parador spain tourist paradores summers end the clan maclean ballad romances first aid for depression 5320b sesog international studies in taxation law and economics selbstmanagement ressourcenorientiert maja storch ebook inncom e528 thermostat avexfx tennessee medicaid manual 2015 wlets the evolution of consciousnessas revealed through the patient has the floor zaraa exemplar for tourism 2014 helano jcb compact tractor service manual abilantis komatsu pc200 6 factory service repair manual ebook | www chemistry essay answers trupin saving energy growing jobs how environmental protection history of the war in hungary in 1848 and 1849 benjay 1935 1942 packard 120 body repair shop manual reprint bouquet of thorns valcap manual nissan quest 1997 billballam cummins 903 service manual coreysmith ford falcon xc workshop manual towies ergonomics foundational principles applications and air sparging for site remediation ebook | dr-calorie
At the Institute for the Learning Sciences we have been developing large scale hypermedia systems... more At the Institute for the Learning Sciences we have been developing large scale hypermedia systems, called ASK systems, that are designed to simulate aspects of conversations with experts. They provide access to manually indexed, multimedia databases of story units. We are particularly concerned with finding a practical solution to the problem of finding indices for thes units when the database grows too large for manual techniques. Our solution is to provide automated assistance that proposes relative links between units, eliminating the need for manual unit-to-unit comparison. In this paper we describe eight classes of links, and show a representation and inference procedure to assist in locating instances of each.
Our research concerns the construction of knowledge-rich memories, in hypermedia form, for use as... more Our research concerns the construction of knowledge-rich memories, in hypermedia form, for use as aids to problem-solving. One of the most difficult steps in building such memories is constructing a rich set of links between the content elements they contain (i.e., text segments, graphics, and video clips). This paper describes point linking, a method we have developed for automated linking. Point linking is currently being incorporated into the ASKTool, a hypermedia editor in ongoing use at the Institute for the Learning Sciences.
Elsevier eBooks, 1989
This chapter provides an overview of how Protos represents concepts, performs classification, and... more This chapter provides an overview of how Protos represents concepts, performs classification, and learns. Protos represents concepts extensionally as collections of exemplars. Accompanying domain knowledge provides cohesiveness within these categories. Classification is a two-step process. The features of a new case provide hypotheses as to its classification. Protos seeks to verify a hypothesis by retrieving an appropriate exemplar and constructing an explanation of its similarity to the new case. When it cannot classify a new case or cannot explain a classification, Protos learns by interacting with its teacher. It selectively retains exemplars and explanations and formulates indices, which allow efficient retrieval during future problem solving. The Protos approach attempts to expand the bandwidth of communication between the teacher and learning system to reduce the amount of training required for concept acquisition. Protos autonomously learns the things that it can reasonably be expected to learn. It depends upon the interaction with the teacher to acquire the knowledge that is outside of the scope of its abilities.
Elsevier eBooks, 1989
This chapter presents the design of Protos, explaining how the Protos approach has been instantia... more This chapter presents the design of Protos, explaining how the Protos approach has been instantiated as a concept representation and a set of algorithms. The chapter elaborates and follows the organization of Chapter 2. First, the representation of Protos' category structure is explained. The featural representation of cases is described, and Protos' explanation language is defined. Second, the classification procedure is explained. Hypothesis formation, exemplar selection, knowledge-based pattern matching, and similarity assessment are described. Third, the details of Protos' learning procedures are presented. The selective retention of exemplars, integration of explanations into domain knowledge, and creation and revision of indices are described. Detailed examples of the classification and learning procedures in action are presented in Chapter 4.