Papers by Kenneth Koedinger
Reifying Implicit Planning in Geometry: Guidelines for Model-Based Intelligent Tutoring System Design
Miss Lindquist: un système fondé sur le dialogue pour apprendre à exprimer algébriquement des énoncés en langage naturel
La symbolisation est la competence permettant de traduire en langage algebrique une situation du ... more La symbolisation est la competence permettant de traduire en langage algebrique une situation du monde reel. A notre sens, c'est la competence principale a acquerir par les eleves pour l'apprentissage de l'algebre au niveau de l'enseignement secondaire. Nous presentons une recherche portant sur les problemes poses par l'acquisition de cette competence, etude qui a mis en evidence une « competence cachee » a l'interieur de la symbolisation. Contrairement a des recherches anterieures qui ont analyse cette acquisition en termes de difficultes de comprehension des textes des enonces et de comprehension du role abstrait des variables, nous pensons que l'algebre est une langue etrangere pour les eleves et que c'est la difficulte pour eux a s'exprimer dans cette langue etrangere qui pose probleme pour l'acquisition de la symbolisation. Nous presentons egalement Mrs Lindquist, un tuteur intelligent (TI) concu pour mener un dialogue pedagogique sur des...
Note-taking, selecting, and choice
Proceedings of the 8th ACM/IEEE-CS joint conference on Digital libraries - JCDL '08, 2008
Our research develops note-taking applications for educational environments. Previous studies fou... more Our research develops note-taking applications for educational environments. Previous studies found that while copy-pasting notes can be more efficient than typing, for some users it reduces attention and learning. This paper presents two studies aimed at designing and evaluating interfaces that encourage focusing. While we were able to produce interfaces that increased desirable behaviors and improved satisfaction, the new interfaces did not improve learning. We suggest design recommendations derived from these studies, and describe a "selecting-to-read" behavior we encountered, which has implications for the design of reading and note-taking applications.
Lecture Notes in Computer Science, 2014
Developing intelligent tutoring systems from student solution data is a promising approach to fac... more Developing intelligent tutoring systems from student solution data is a promising approach to facilitating more widespread application of tutors. In principle, tutor feedback can be generated by matching student solution attempts to stored intermediate solution states, and next-step hints can be generated by finding a path from a student's current state to a correct solution state. However, exact matching of states and paths does not work for many domains, like programming, where the number of solution states and paths is too large to cover with data. It has previously been demonstrated that the state space can be substantially reduced using canonicalizing operations that abstract states. In this paper, we show how solution paths can be constructed from these abstract states that go beyond the paths directly observed in the data. We describe a domain-independent algorithm that can automate hint generation through use of these paths. Through path construction, less data is needed for more complete hint generation. We provide examples of hints generated by this algorithm in the domain of programming.
Lecture Notes in Computer Science, 2012
Intelligent Tutoring Systems, 2002
Research in machine learning is making it possible for instructional developers to perform format... more Research in machine learning is making it possible for instructional developers to perform formative evaluations of different curricula using simulated students (VanLehn, Ohlsson & Nason, 1993). Experiments using simulated students can help clarify issues of instructional design, such as when a complex skill can be better learned by being broken into components. This paper describes two formative evaluations using simulated students that shed light on the potential benefits and limitations of mastery learning. Using an ACT-R based cognitive model (Anderson & Lebiere, 1998) we show that while mastery learning can contribute to success in some cases (Corbett & Anderson, 1995), it may actually impede learning in others. Mastery learning was crucial to learning success in an experiment comparing a traditional early algebra curriculum to a novel one presenting verbal problems first. However, in a second experiment, an instructional manipulation that contradicts mastery learning led to greater success than one consistent with it. In that experiment learning was better when more difficult problems were inserted earlier in the instructional sequence. Such problems are more difficult not because they have more components but because they cannot be successfully solved using shallow procedures that work on easier problems.
Open-Ended Tutoring for Programming: Building Next-Step Hints into an Online Development Environment
Automatically generated feedback could improve the learning gains of novice programmers, especial... more Automatically generated feedback could improve the learning gains of novice programmers, especially for students who are in large classes where instructor time is limited. We propose a data-driven approach for automatic feedback generation which utilizes the program solution space to predict where a student is located within the set of many possible learning progressions and what their next steps should be. This paper describes the work we have done in implementing this approach and the challenges which arise when supporting ill-defined domains.
Proceedings of the 2004 conference on Human factors in computing systems - CHI '04, 2004
Although engineering models of user behavior have enjoyed a rich history in HCI, they have yet to... more Although engineering models of user behavior have enjoyed a rich history in HCI, they have yet to have a widespread impact due to the complexities of the modeling process. In this paper we describe a development system in which designers generate predictive cognitive models of user behavior simply by demonstrating tasks on HTML mock-ups of new interfaces. Keystroke-Level Models are produced automatically using new rules for placing mental operators, then implemented in the ACT-R cognitive architecture. They interact with the mock-up through integrated perceptual and motor modules, generating behavior that is automatically quantified and easily examined. Using a query-entry user interface as an example [19], we demonstrate that this new system enables more rapid development of predictive models, with more accurate results, than previously published models of these tasks.
Lecture Notes in Computer Science, 1998
We present a vision for learning environments, called Science Learning Spaces, that are rich in e... more We present a vision for learning environments, called Science Learning Spaces, that are rich in engaging content and activities, provide constructive experiences in scientific process skills, and are as instructionally effective as a personal tutor. A Science Learning Space combines three independent software systems: 1) lab/field simulations in which experiments are run and data is collected, 2) modeling/construction tools in which data representations are created, analyzed and presented, and 3) tutor agents that provide just-in-time assistance in higher order skills like experimental strategy, representational tool choice, conjecturing, and argument. We believe that achieving this ambitious vision will require collaborative efforts facilitated by a component-based software architecture. We have created a feasibility demonstration that serves as an example and a call for further work toward achieving this vision. In our demonstration, we combined 1) the Active Illustrations lab simulation environment, 2) the Belvedere argumentation environment, and 3) a modeltracing Experimentation Tutor Agent. We illustrate student interaction in this Learning Space and discuss the requirements, advantages, and challenges in creating one.
CHI '10 Extended Abstracts on Human Factors in Computing Systems, 2010
Pen-based Flash Cards Application ("application") offers the flexibility of handwritten input whi... more Pen-based Flash Cards Application ("application") offers the flexibility of handwritten input while benefiting a wide set of users to increase their memory retention. It is particularly useful in learning mathematics where typing the material using a keyboard can be difficult. In this study, we describe the observations and major findings in a two-year case study in an eighth-grade geometry class. We found that this application may enhance teacher-student interaction, increase autonomy in students for self-guided learning, and encourage collaborative learning.
Tangible Collaborative Learning with a Mixed-Reality Game: EarthShake
Lecture Notes in Computer Science, 2013
We explore the potential of bringing together the advantages of computer games and the physical w... more We explore the potential of bringing together the advantages of computer games and the physical world to increase engagement, collaboration and learning. We introduce EarthShake: A tangible interface and mixed-reality game consisting of an interactive multimodal earthquake table, block towers and a computer game synchronized with the physical world via depth camera sensing. EarthShake helps kids discover physics principles while experimenting with real blocks in a physical environment supported with audio and visual feedback. Students interactively make predictions, see results, grapple with disconfirming evidence and formulate explanations in forms of general principles. We report on a preliminary user study with 12 children, ages 4-8, indicating that EarthShake produces large and significant learning gains, improvement in explanation of physics concepts, and clear signs of productive collaboration and high engagement.
Selection-based note-taking applications
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2007
ABSTRACT The increasing integration of education and technology has led,to the,development,of,a,r... more ABSTRACT The increasing integration of education and technology has led,to the,development,of,a,range,of,note-taking applications. Our project’s goal is to provide empirical data to guide the design of such note-taking applications by evaluating,the,behavioral,and,learning,outcomes,of different note-taking functionality. The study reported here compares,note-taking using,a,text editor and,four interaction techniques. The two standard techniques are typing and,copy-paste. The two,novel,techniques,are restricted copy-paste and menu-selection, intended to increase,attention,and,processing,respectively. Hypothesized,learning
Evaluating the effect of technology on note-taking and learning
CHI '06 Extended Abstracts on Human Factors in Computing Systems, 2006
... ACM 1-59593-298-4/06/0004. Aaron Bauer Human-Computer Interaction Institute 5000 Forbes Ave P... more ... ACM 1-59593-298-4/06/0004. Aaron Bauer Human-Computer Interaction Institute 5000 Forbes Ave Pittsburgh, PA, 15213 abauer@cmu.edu Kenneth Koedinger Human-Computer Interaction Institute 5000 Forbes Ave Pittsburgh, PA, 15213 koedinger@cmu.edu ...
User Modeling 2003, 2003
Numerous approaches to student modeling have been proposed since the inception of the field more ... more Numerous approaches to student modeling have been proposed since the inception of the field more than three decades ago. hat the field is lacking completely is comparative analyses of different student modeling approaches. Such analyses are sorely needed, as they can identify the most promising approaches and provide guidelines for future research. In this paper we compare Cognitive Tutoring to Constraint-Based Modeling (CBM). We present our experiences in implementing a database design tutor using both methodologies and highlight their strengths and weaknesses. We compare their characteristics and argue the differences are often more apparent than real. For specific domains, one approach may be favoured over the other, making them viable complementary methods for supporting learning.
Sixth IEEE International Conference on Advanced Learning Technologies (ICALT'06)
Positioning in learning networks is a process that assists learners in finding a starting point a... more Positioning in learning networks is a process that assists learners in finding a starting point and an efficient route in the network that will foster competence building. In order to avoid labor-intense routines in the network we explore computational approaches to services such as positioning that are based on the contents of the learning network and the behavior of those participating in it, rather than in predefined procedures and (meta-) data. We present a content-based approach to positioning that uses latent semantic analysis to compare the learner's portfolio to the contents offered in the learning network. Although initial results indicate the feasibility of the approach there are a number of important caveats to consider
Intelligent Tutoring Systems, 2006
To implement real intelligence or adaptivity, the models for intelligent tutoring systems should ... more To implement real intelligence or adaptivity, the models for intelligent tutoring systems should be learnt from data. However, the educational data sets are so small that machine learning methods cannot be applied directly. In this paper, we tackle this problem, and give general outlines for creating accurate classifiers for educational data. We describe our experiment, where we were able to predict course success with more than 80% accuracy in the middle of course, given only hundred rows of data.
Cognitive tutors as modeling tool and instructional model
User Modeling and User-Adapted Interaction, 2010
Personalised environments such as adaptive educational systems can be evaluated and compared usin... more Personalised environments such as adaptive educational systems can be evaluated and compared using performance curves. Such summative studies are useful for determining whether or not new modifications enhance or degrade performance. Performance curves also have the potential to be utilised in formative studies that can shape adaptive model design at a much finer level of granularity. We describe the use of learning curves for evaluating personalised educational systems and outline some of the potential pitfalls and how they may be overcome. We then describe three studies in which we demonstrate how learning curves can be used to drive changes in the user model. First, we show how using learning curves for subsets of the domain model can yield insight into the appropriateness of the model's structure. In the second study we use this method to experiment with model granularity. Finally, we use learning curves to analyse a large volume of user data to explore the feasibility of using them as a reliable method for finetuning a system's model. The results of these experiments demonstrate the successful use of performance curves in formative studies of adaptive educational systems.
Student Modeling in Intelligent Tutoring Systems: Acquisition of Cognitive Skill and Tutorial Interventions
Social Science Computer Review, 1993
In intelligent tutoring systems (ITS) student models are used for diagnosis resulting in didactic... more In intelligent tutoring systems (ITS) student models are used for diagnosis resulting in didactical choices. Among others, didactical choices concern at a global level decisions about the sequencing of learning materials and at a local level decisions about presenting additional material and remediation. In ITS didactical choices are often implicit. In some cases choices are derived from models of human tutoring behavior. As far as ITS are based on student (and expert) models, which are plausible from the cognitive psychological point of view, implicit didactical choices or human tutor models can be avoided. When a student model is constructed according to the principles of ACT*, predictions of the effects of didactical interventions on learning outcomes can be made. It can be shown that learning outcomes as speed of processing and probability of correct response are mediated by the storage-and-strengthening parameters of the control structure. Storage and strengthemng in turn are i...
Uploads
Papers by Kenneth Koedinger