Joseph Beck, Mia Stern, and Beverly Park Woolf. Using the student model to control problem difficulty. In Anthony Jameson, Ccile Paris, and Carlo Tasso, editors, User Modeling: Proceedings of the Sixth International Conference, UM97, pages 277--288. Springer, Vienna, New York, 1997. Available from http://www.um.org.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....However, adaptation might still be unsuccessful because users become confused or dissatisfied. Thus, the human system interaction has to be evaluated as well. Both, objective and subjective measures are relevant. e.g. users might rate the system s usability [7, 14] or the solution quality [1]. Examples of objective criteria for interaction quality include frequency of task success and number of required hints [10] The examples above emphasize the necessity of empirical evaluations in each of the four layers. It is impossible to detect certain kinds of mis adaptations that result ....

Joseph Beck, Mia Stern, and Beverly Park Woolf. Using the student model to control problem difficulty. In Anthony Jameson, Ccile Paris, and Carlo Tasso, editors, User Modeling: Proceedings of the Sixth International Conference, UM97, pages 277--288. Springer, Vienna, New York, 1997. Available from http://www.um.org.

.... the BSS1 tutoring system a general fuzzy logic engine was designed and implemented to support development of intelligent features which can better manage the student s learning [23] Uncertainty of student s performance in Sherlock II and in the MDF tutor was managed with fuzzy distributions [5] [2]. For a tutoring system in the domain of physics the Knowledge and Learning Student Model was designed using fuzzy logic techniques, inferring about student s knowledge level and cognitive abilities from student s behavior [14] 15] Neural networks have also been used in the design of ITSs, ....

Beck,J Stern,M and Woolf, B,P. Using the Student Model to Control Problem Difficulty. World wide web document

....by users and to the language in which these activities are expressed. The emphasis which VIS places upon extending the learner beyond what she can achieve alone and then providing sufficient assistance to ensure that she does not fail also sets it apart from other system s such as that of Beck, Stern and Woolf (1997), which generate problems of controlled difficulty and aim to tailor the hints and help the system offers to the individual s particular needs. VIS extends the work done with other systems which have used the ZPD concept in the learner modelling (e.g. Gegg Harrison, 1992) 2 VIS and the Ecolab ....

Beck, J., Stern, M., & Woolf, B. P. (1997). Using the student model to control problem difficulty. In A. Jameson, C. Paris, & C. Tasso (Eds.), User Modeling: Proceedings of Sixth International Conferenceon User Modeling, UM97. New York: Springer Wien. 278-288.

....Operators include addition, subtraction, multiplication, and division. Operands are wholes and fractions. There are also topics on pre fraction skills that do not fit this hierarchy well. Each topic may have a number of subskills which are optional steps that may be needed to solve a problem[4]. For example, simplifying the result is a subskill of add fractions. Subskills have a level of difficulty associated with them. Borrowing across several columns in a subtraction problem is more difficult than borrowing from the column directly to the left. The tutor provides feedback when ....

....minutes. Feature type Information Student Gender Performance in 9 tasks of Piagetian level of cognitive development[2] Combined score for Piaget tasks Proficiency at current topic Topic Operator (one hot encoding) Operand (one hot encoding) Problem Problem difficulty[4] Size of operands answer Number of subskills required[4] Difficulty of subskills Context Is this student s first attempt Number of prior mistakes Best hint seen Current hint depth (hints can create subhints) Maximum hint depth on this problem Best hint seen ....

[Article contains additional citation context not shown here]

J. E. Beck, M. Stern, and B.P. Woolf. Using the student model to control problem difficulty. In Proceedings of the Seventh International Conference on User Modeling, pages 277--288, 1997.

....it must maintain an accurate assessment of the stu dent s strengths and weaknesses in this task domain. Online self assessment surveys conducted as students work with WhaleWatch have shown that the tutor generates a more accurate assessment of each student s abilities than the students themselves (Beck, et al. 1997a) WhaleWatch uses Artificial Intelligence techniques for problem generation, hint selection and student modeling (Beck, et al. 1997b; Beal, et al. 1998) Multimedia is used judiciously to engage the student by animating key concepts and providing interactive manipulables based on those used by ....

.... conducted as students work with WhaleWatch have shown that the tutor generates a more accurate assessment of each student s abilities than the students themselves (Beck, et al. 1997a) WhaleWatch uses Artificial Intelligence techniques for problem generation, hint selection and student modeling (Beck, et al. 1997b; Beal, et al. 1998) Multimedia is used judiciously to engage the student by animating key concepts and providing interactive manipulables based on those used by classroom teachers. WhaleWatch Testing and Results Classroom trials of the WhaleWatch prototype for purposes of formative ....

[Article contains additional citation context not shown here]

Beck, J.E, Stern, M., & Woolf, B.P. (1997b). Using the Student Model to Control Problem Difficulty. In Proceedings of the Seventh International Conference on User Modeling, 1997. 277-288.

....instruction and supportive feedback to each student. This individualized instruction is provided by adjusting the level of feedback, controlling the progression through the curriculum, and constructing problems at the appropriate level of difficulty. A description of this process can be found in [2]. By making sure the instruction is appropriate for each student, we ensure that students do not become overly discouraged. We hypothesized that this feature should be particularly effective with female students, who are more easily discouraged about their progress in mathematics. 2 Experiments ....

Beck, J. E., Stern, M., & Woolf, B.P. (1997). Using the student model to control problem difficulty. In Proceedings of the Seventh International Conference on User Modeling, pp. 277-288.

....This state is composed of 48 features in four main areas: 1. Student: The student s level of prior proficiency and level of cognitive development(Arroyo et al. 2000) 2. Topic: How hard the current topic is and the type of operand operators. 3. Problem: How complex is the current problem (Beck, Stern, Woolf 1997). 4. Context: Describes the student s current efforts at answering this question, and hints he has seen. High level student modeling Most student models are concerned with representing the student s level of ability on distinct portions of the domain. Although useful, it is not always obvious ....

....that we are able to include types of data not normally found in an ITS. For instance, we do not have a good theory for how level of cognitive development should impact hint selection. How then should such data be added to the model Prior research has come across this same difficulty(Shute 1995; Beck, Stern, Woolf 1997). By automating model construction, we bypass this issue entirely. Also, our model of high level performance is executable. Both previous mistakes and time spent on the problem are part of the context features maintained about a problem. Therefore, the model can predict the student s probability ....

Beck, J. E.; Stern, M.; and Woolf, B. 1997. Using the student model to control problem difficulty. In Proceedings of the Seventh International Conference on User Modeling, 277--288.

....fifth and sixth graders. This system adapts its instruction to meet the needs of each learner by intelligently selecting a topic on which the student should work, providing hints that match the student s level of ability, and dynamically constructing problems that are appropriate for the student [2]. The system represents the domain as a topic network. A topic refers to a large unit of knowledge, such as subtract fractions or multiply whole numbers. Each topic has a list of pretopics; before the student can work on a topic, all of its pretopics must be passed. In addition, the system ....

....learning techniques described here were added post hoc, and have been applied to the data gathered during the evaluation. We are working to incorporate these mechanisms into the next version of the tutor. Currently, the MFD tutor adjusts problem difficulty via a complex, but ad hoc mechanism [2]. The system rates a problem s complexity on an absolute scale, builds problems that are at a good level of difficulty for each student and require the student to apply skills in which she needs more practice. Rather than controlling the problem s complexity directly, the new system will ....

Beck, J. and Stern, M. and Woolf, B.P.: Using the Student Model to Control Problem Difficulty. In: Proceedings of the Seventh International Conference on User Modeling. (1997) 277--288

....the example space that comes closest to the actual behavior for the examples for this user, and use that example space for the user. If there is not an already existing example space that is close enough , then a new space is created for this user. 2. 3 Using the technique in MFD The MFD tutor [1] is designed to teach grade school students fraction arithmetic. In the present version of the system, teaching strategies are predefined. The next version will use machine learning techniques to determine how to customize teaching for each student. 2.3.1 The prediction task Our goal is to use a ....

J. Beck, M. Stern, and B.P. Woolf. Using the Student Model to Control Problem Difficulty. In Proceedings of the Sixth International Conference on User Modeling, pages 277--289, 1997.

....pretest information that could be incorporated into the system) of the variance in a posttest. This evidence supports further investigation into using regression techniques to update student models. 3. 2 System background This framework is being tested in MFD (Mixed numbers, Fractions, Decimals) [2], a tutor that teaches fraction arithmetic to grade school children. The current system is somewhat different than the previous version [9] for community college students in that it performs more teaching as opposed to primarily supporting problem solving. It retains many of the same design ....

....tested with 19 fifth grade children in a rural school. Students used the tutor for 6 sessions over a period of two weeks. On average, each student completed about 25 problems. The students did not work on the same problems; the system created problems to meet the student s current level of ability [2]. When the student submits an answer, simply asking for his overall confidence that his answer is correct does not suffice. He may be unconfident because solving the problem required the use of an unfamiliar substep, but in general he could solve problems of this type. What skill should the ....

[Article contains additional citation context not shown here]

Beck, J., Stern, M. and Woolf, B. 1997. Using the Student Model to Control Problem Difficulty. To Appear in Proceedings of the Sixth International Conference on User Modeling.

....measure the degree of the student s ability in a topic . This measurement commonly takes the form of the probability a student has mastered a topic. This suffices for basic reasoning such as whether to promote the student to a more complex topic, or to control the activity s level of difficulty [1]. However, frequently more complex types of reasoning are needed. For example, should the system review a previously learned topic, should it give the student a problem to solve, or should it first present another example of the process involved Ideally, the answers to such questions would be ....

....features. Fortunately, much progress has been made on the former via Bayesian Networks [2] and other formal reasoning techniques while less progress has been made on the latter. 2 Motivation This framework is being designed for use in the MFD (Mixed Number, Fractions, Decimals) system [1]. The goal of this system is to both to help grade school students learn fractional and decimal arithmetic, and to increase the confidence of students using the system. This system has some heuristics that help determine if a student needs remediation in a topic, has forgotten a topic, needs to ....

J. Beck, M. Stern, and B. Woolf. Using the student model to control problem difficulty. In Proceedings of the Seventh International Conference on User Modeling, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC