Student Modeling

Artificial Intelligence applications in educational field are getting more and more popular during the last decade (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009) and that is why much relevant research has been conducted. In this paper, we present the most interesting attempts to apply artificial intelligence methods such as fuzzy logic, neural networks, genetic programming and hybrid approaches such as neurofuzzy systems and genetic programming neural networks (GPNN) in student modeling. This latest research trend is a part of every Intelligent Tutoring System and aims at generating and updating a student model in order to modify learning content to fit individual needs or to provide reliable assessment and feedback to student's answers. In this paper, we make a brief presentation of methods used to point out their qualities and then we attempt a navigation to the most representative studies sought in the decade of our interest after classifying them according to the principal aim they attempted to serve.

Artificial Intelligence applications in educational field are getting more and more popular during the last decade (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009) and that is why much relevant research has been conducted. In this paper, we present the most interesting attempts to apply artificial intelligence methods such as fuzzy logic, neural networks, genetic programming and hybrid approaches such as neurofuzzy systems and genetic programming neural networks (GPNN) in student modeling. This latest research trend is a part of every Intelligent Tutoring System and aims at generating and updating a student model in order to modify learning content to fit individual needs or to provide reliable assessment and feedback to student's answers. In this paper, we make a brief presentation of methods used to point out their qualities and then we attempt a navigation to the most representative studies sought in the decade of our interest after classifying them according to the principal aim they attempted to serve.

Open learner models (OLM) can support self-regulated learning, collaborative interaction, and navigation in adaptive educational systems. Previous research has found that learners have a range of preferences for learner model visualisation. However, research has focused mainly on visualisations that are available in a single system, meaning that not all visualisations have been compared to each other. We present a study using screen shots of OLM visualisations for individuals and for comparing one's own learner model to the models of other individuals or the group, to define visualisations that students would be able to use to identify their next steps, across a wider range of options.

Computer assisted learning is a common issue among the research communities globally. The benefits of such technologies are widely accepted, so more and more educational applications are being developed. However, educational software aims to be used by students for learning. Therefore educational applications have to be as attractive as possible to increase the engagement of students. To serve this purpose many researchers employ several aspects of multimedia technology to improve the aesthetics and the appeal of educational software. One very attractive and popular application area, that involves and demands advanced multimedia technologies and could be used for educational purposes, is computer games. In this paper a virtual reality game for educational purposes is described. This tutoring system incorporates a virtual reality user interface similar to that of common commercial games. The game is enriched with student modeling mechanisms that ensure the individualisation of the in...

Ubiquitous learning allows students to learn at any time and any place. Adaptivity plays an important role in ubiquitous learning, aiming at providing students with adaptive and personalized learning material, activities, and information at the right place and the right time. However, for providing rich adaptivity, the student model needs to be able to gather a variety of information about the students. In this paper, an automatic, global, and dynamic student modeling approach is introduced, which aims at identifying and frequently updating information about students’ progress, learning styles, interests and knowledge level, problem solving abilities, preferences for using the system, social connectivity, and current location. This information is gathered in an automatic way, using students’ behavior and actions in different learning situations provided by different components/services of the ubiquitous learning environment. By providing a comprehensive student model, students can be supported by rich adaptivity in every component/service of the learning environment. Furthermore, the information in the student model can help in giving teachers a better understanding about the students’ learning process.

In this paper, we describe the integration and evaluation of an existing generic Bayesian student model (GBSM) into an existing computerized testing system within the Mathematics Education Project (PmatE-Projecto Matemática Ensino) of the University of Aveiro. This generic Bayesian student model was previously evaluated with simulated students, but a real application was still missing. In the work presented here, we have used the GBSM to define Bayesian Student Models (BSMs) for a concrete domain: first degree equations. In order to test the diagnosis capabilities of such BSMs, an evaluation with 152 students has been performed. Each of the 152 students took both a computerized test within PMatE and a written exam, both of them designed to measure students knowledge in 12 concepts related to first degree equations. The written exam was graded by three experts. Then two BSMs were developed, one for the computer test and another one for the written exam. These BSMs were used to to obtain estimations of student's knowledge on the same 12 concepts, and the inter-rater agreement among the different measures was computed. Results show a high degree of agreement among the scores given by the experts and also among the diagnosis provided by the BSM in the written exam and the experts average, but a low degree of agreement among the diagnosis provided by the BSM in the computer test and expert's average.

We developed a new model for iList, our system that helps students learn linked list. The model is automatically extracted from past student data, and allows iList to track students' problem-solving behavior in order to provide targeted feedback. We evaluated the new model both intrinsically and extrinsically. We show that the model can match most student actions after a relatively small sequence of observations, and that iList can effectively use the new student tracker to provide feedback and help students learn.

Thanks to the exponential growth of the Internet, Distance Education is becoming more and more strategic in many fields of daily life. Its main advantage is that students can learn through appropriate web platforms that allow them to take advantage of multimedia and interactive teaching materials, without constraints neither of time nor of space. Today, in fact, the Internet offers many platforms suitable for this purpose, such as Moodle, ATutor and others. Coursera is another example of a platform that offers different courses to thousands of enrolled students. This approach to learning is, however, posing new problems such as that of the assessment of the learning status of the learner in the case where there were thousands of students following a course, as is in Massive On-line Courses (MOOC). The Peer Assessment can therefore be a solution to this problem: evaluation takes place between peers, creating a dynamic in the community of learners that evolves autonomously. In this article, we present a first step towards this direction through a peer assessment mechanism led by the teacher who intervenes by evaluating a very small part of the students. Through a mechanism based on machine learning, and in particular on a modified form of K-NN, given the teacher's grades, the system should converge towards an evaluation that is as similar as possible to the one that the teacher would have given. An experiment is presented with encouraging results.

PurposeInterest is currently growing in open social learner modeling (OSLM), which means making peer models and a learner's own model visible to encourage users in e-learning. The purpose of this study is to examine students' views about the OSLM in an e-learning system.Design/methodology/approachThis case study was conducted with 40 undergraduate students enrolled in advanced programming and database management system courses. A Likert-type questionnaire and open-ended questions were used to obtain the students' views. System usage data were also analyzed to ensure the richness and diversity of the overall data set.FindingsThe quantitative data of the students' views were analyzed with descriptive statistics; the results are presented as graphics. The qualitative data of the students' views were examined by content analysis to derive themes. These themes are organized into four subtopics: the students' positive views, their negative views, their improvement ...

The ICICLE project is a Computer-Assisted Language Learning (CALL) environment geared toward teaching English as a second language. This paper reports on an initial prototype application of the system for deaf learners of written English. A primary focus of the ICICLE effort has been devoted to enabling the system to adapt to a learning user over the evolution of language proficiency. In this paper, we overview and motivate the design of our novel user modeling component that integrates Selinker's Interlanguage theory and other research in Second Language Acquisition in order to accurately represent a learner's internal grammar as it changes over time. The objectives of this effort are twofold: to accurately diagnose and respond to learner errors, and to focus tutorial feedback on those errors which are most relevant to the learner's acquisition process.

Learning styles and affective states have a significant effect on student learning. The aim of this paper is to present a concept to identify and integrate learning styles and affective states of a learner into web-based learning management systems and therefore providing learners with adaptive courses and additional individualized pedagogical guidance that is tailored to their learning styles and affective states. Through considering affective states and learning styles, learners are provided with a learning environment that is more personalized and tailored to learners needs and current situation, leading to better learning outcomes and progress.

In this paper, we describe the integration and evaluation of an existing generic Bayesian student model (GBSM) into an existing computerized testing system within the Mathematics Education Project (PmatE-Projecto Matemática Ensino) of the University of Aveiro. This generic Bayesian student model was previously evaluated with simulated students, but a real application was still missing. In the work presented here, we have used the GBSM to define Bayesian Student Models (BSMs) for a concrete domain: first degree equations. In order to test the diagnosis capabilities of such BSMs, an evaluation with 152 students has been performed. Each of the 152 students took both a computerized test within PMatE and a written exam, both of them designed to measure students knowledge in 12 concepts related to first degree equations. The written exam was graded by three experts. Then two BSMs were developed, one for the computer test and another one for the written exam. These BSMs were used to to obtain estimations of student's knowledge on the same 12 concepts, and the inter-rater agreement among the different measures was computed. Results show a high degree of agreement among the scores given by the experts and also among the diagnosis provided by the BSM in the written exam and the experts average, but a low degree of agreement among the diagnosis provided by the BSM in the computer test and expert's average.

This paper presents an ontology-based methodology for automatic decomposition of Learning Objects (LOs) into reusable content units, and discusses their dynamic assembly into personalized learning paths within the domain of technology-assisted self-directed learning. PerLE, a learner- centered, adaptive, tutored Personal Learning Environment (PLE), was developed to substantiate our dynamic syllabus approach applicable to authoring tutored adaptive e-courses. PerLE allows the decomposition of LOs into smaller learning units, which can be dynamically assembled into new LO sequences and repurposed for different learning objectives. While focusing on ontologies in the context of user modeling and personalization, we particularly describe the concept of creating dynamically assembled e-course sequences. We describe how PerLE was designed to better respond to psychological issues of self-directed learning through the strategic approach of user profiling, grounded in the Felder- Silverman ...

PurposeInterest is currently growing in open social learner modeling (OSLM), which means making peer models and a learner's own model visible to encourage users in e-learning. The purpose of this study is to examine students' views about the OSLM in an e-learning system.Design/methodology/approachThis case study was conducted with 40 undergraduate students enrolled in advanced programming and database management system courses. A Likert-type questionnaire and open-ended questions were used to obtain the students' views. System usage data were also analyzed to ensure the richness and diversity of the overall data set.FindingsThe quantitative data of the students' views were analyzed with descriptive statistics; the results are presented as graphics. The qualitative data of the students' views were examined by content analysis to derive themes. These themes are organized into four subtopics: the students' positive views, their negative views, their improvement ...

This paper presents an experimental evaluation of eye gaze data as a source for modeling user's learning in Interactive Simulations (IS). We compare the performance of classifier user models trained only on gaze data vs. models trained only on interface actions vs. models trained on the combination of these two sources of user interaction data. Our long-term goal is to build user models that can trigger adaptive support for students who do not learn well with ISs, caused by the often unstructured and open-ended nature of these environments. The test-bed for our work is the CSP applet, an IS for Constraint Satisfaction Problems (CSP). Our findings show that including gaze data as an additional source of information to the CSP applet's user model significantly improves model accuracy compared to using interface actions or gaze data alone.

Considering the increasing importance of adaptive approaches in CALL systems, this study implemented a machine learning based student modeling middleware with Bayesian networks. The profiling approach of the student modeling system is based on Felder and Silverman's Learning Styles Model and Felder and Soloman's Index of Learning Styles Questionnaire. The questionnaire was adapted to Turkish for this experimental study conducted with respect to the visual/verbal and active/reflective dimensions of the model. A topic in EFL was chosen for the learning content design, which was also carried into the digital domain and remastered as separate learning scenes for different learning styles. Computer software was also implemented to carry out the experimental learning processes. A quasi-experimental pre-test, post-test design was conducted with 46 volunteers, with 23 students assigned each to a control and an experimental group to compare academic achievement between student-based learning and conventional computer-based learning. No significant difference was found in academic achievement between the control and experimental groups after the experimental treatment. The diagnostic performance of the proposed student modeling system was also compared with performances from similar studies. This student modeling system had a successful prediction rate of 41% on the visual/verbal dimension and 54% on the active/reflective dimension, respectively.

Bayesian Knowledge Tracing (BKT)[1] is a user modeling method extensively used in the area of Intelligent Tutoring Systems. In the standard BKT implementation, there are only skill-specific parameters. However, a large body of research strongly suggests that studentspecific variability in the data, when accounted for, could enhance model accuracy [5, 6, 8]. In this work, we revisit the problem of introducing student-specific parameters into BKT on a larger scale. We show that student-specific parameters lead to a tangible improvement when predicting the data of unseen students, and that parameterizing students' speed of learning is more beneficial than parameterizing a priori knowledge.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

In a Exploratory Learning Environment users acquire knowledge while freely experiencing the environment. In this setting, it is often hard to identify actions or behaviors as correct or faulty, making it hard to provide adaptive support to students who do not learn well with these environments. In this paper we discuss an approach that uses Class Association Rule mining and a Class Association Rule Classifier to identify relevant interaction patterns and build student models for online classification. We apply the approach to generate a student model for an ELE for AI algorithms and present preliminary results on its effectiveness

In a Exploratory Learning Environment users acquire knowledge while freely experiencing the environment. In this setting, it is often hard to identify actions or behaviors as correct or faulty, making it hard to provide adaptive support to students who do not learn well with these environments. In this paper we discuss an approach that uses Class Association Rule mining and a Class Association Rule Classifier to identify relevant interaction patterns and build student models for online classification. We apply the approach to generate a student model for an ELE for AI algorithms and present preliminary results on its effectiveness

This paper suggests different methods for visualising uncertainty in open learner models (OLM). In order to visualise the uncertainty in OLMs, two factors need to be measured, namely the source of the uncertainty in the data and the level of uncertainty in the learner model. This paper proposes a method to detect the source of uncertainty within a learner model: outlier analysis is employed to identify inconsistencies in the data set from which the OLM is built. The level of uncertainty that is present in the model is determined by summing the influence weights of the learner model data that was identified as being inconsistent. Different approaches to visualising this uncertainty within OLMs are proposed; and benefits for OLMs that visualise uncertainty in learner models that can be jointly maintained by student and system, are argued.

Intelligent tutoring systems that utilize Bayesian Knowledge Tracing have achieved the ability to accurately predict student performance not only within the intelligent tutoring system, but on paper post-tests outside of the system. Recent work has suggested that contextual estimation of student guessing and slipping leads to better prediction within the tutoring software (Baker, Corbett, & Aleven, 2008a, 2008b). However, it is not yet clear whether this new variant on knowledge tracing is effective at predicting the latent student knowledge that leads to successful post-test performance. In this paper, we compare the Contextual-Guess-and-Slip variant on Bayesian Knowledge Tracing to classical four-parameter Bayesian Knowledge Tracing and the Individual Difference Weights variant of Bayesian Knowledge Tracing (Corbett & Anderson, 1995), investigating how well each model variant predicts post-test performance. We also test other ways to utilize contextual estimation of slipping withi...

Abstract. In this paper a question generation approach for adaptive assessment is purposed to estimate the student knowledge model in a probabilistic domain within an intelligent tutoring system. Assessing questions are generated adaptively according to the student knowledge based on two factors (i) the student misconceptions that are entailed in the student knowledge model, and (ii) the information gain maximizing. Updating and verification of the student model are conducted based on the matching between the student’s and model answers to assessing questions. Comparison between using the adapted questions and random questions is investigated. Results suggest that utilizing adapted generated questions increases the approximation accuracy of the student model by 40 % in addition to decreasing of the required assessing questions by 50 % compared to using fixed questions.

We developed a new model for iList, our system that helps students learn linked list. The model is automatically extracted from past student data, and allows iList to track students' problem-solving behavior in order to provide targeted feedback. We evaluated the new model both intrinsically and extrinsically. We show that the model can match most student actions after a relatively small sequence of observations, and that iList can effectively use the new student tracker to provide feedback and help students learn.

Traditional Knowledge Tracing, which traces students’ knowledge of each decomposed individual skill, has been a popular student model for adaptive tutoring. Unfortunately, such a model fails to model complex skill practices where simple decompositions cannot capture potential additional skills that underlie the context as a whole constituting an interconnected chunk. In this work, we propose a datadriven approach to extract and model potential chunk units in a Knowledge Tracing framework for tracing deeper knowledge, which is primarily based on Bayesian network techniques. We argue that traditional prediction metrics are unable to provide a “deep” evaluation for such student models, and propose novel data-driven evaluations combined with classroom studies in order to examine our proposed student model’s real-world impact on students’ learning.

© 2016 International Educational Data Mining Society. All rights reserved. We propose a novel tensor factorization approach, Feedback-Driven Tensor Factorization (FDTF), for modeling student learning process and predicting student performance. This approach decomposes a tensor that is built upon students’ attempt sequence, while considering the quizzes students select to work with as its feedback. FDTF does not require any prior domain knowledge, such as learning resource skills, concept maps, or Q-matrices. The proposed approach differs significantly from other tensor factorization approaches, as it explicitly models the learning progress of students while interacting with the learning resources. We compare our approach to other state-of-the-art approaches in the task of Predicting Student Performance (PSP). Our experiments show that FDTF performs significantly better compared to baseline methods, including Bayesian Knowledge Tracing and a state-of-the-art tensor factorization appr...

This paper presents an in-depth analysis of a nonconventional topic-based personalization approach for adaptive educational systems (AES) that we have explored for a number of years in the context of university programming courses. With this approach both student modeling and adaptation are based on coarse-grained knowledge units that we called topics. Our motivation for the topic-based personalization was to enhance AES transparency for both teachers and students by utilizing typical topic-based course structures as the foundation for designing all aspects of an AES from the domain model to the end-user interface. We illustrate the details of the topic-based personalization technology, with the help of the Webbased educational service QuizGuide-the first system to implement it. QuizGuide applies the topicbased personalization to guide students to the right learning material in the context of an undergraduate C programming course. While having a number of architectural and practical advantages, the suggested coarse-grained personalization approach deviates from the common practices toward knowledge modeling in AES. Therefore, we believed, that several aspects of QuizGuide required a detailed evaluation-from modeling accuracy to the effectiveness of adaptation. How can one evaluate the soundness of a user modeling (UM) approach, rather than a specific personalized system? The paper discusses how this new student modeling approach can be evaluated, and presents our attempts to evaluate it from multiple different prospects. The evaluation of QuizGuide across several consecutive semesters demonstrates that, although topics do not always support precise UM, they can provide a basis for successful personalization in AESs.

This paper suggests different methods for visualising uncertainty in open learner models (OLM). In order to visualise the uncertainty in OLMs, two factors need to be measured, namely the source of the uncertainty in the data and the level of uncertainty in the learner model. This paper proposes a method to detect the source of uncertainty within a learner model: outlier analysis is employed to identify inconsistencies in the data set from which the OLM is built. The level of uncertainty that is present in the model is determined by summing the influence weights of the learner model data that was identified as being inconsistent. Different approaches to visualising this uncertainty within OLMs are proposed; and benefits for OLMs that visualise uncertainty in learner models that can be jointly maintained by student and system, are argued.

Open learner models to facilitate reflection are becoming more common in adaptive learning environments. There are a variety of approaches to presenting the learner model to the student, and for the student to interact with their open learner model, as the requirements ...