Search

This chapter attempts to untangle the relationships between personalized guidance and domain modeling, as well as to explain how domain modeling could be used to provide personalized guidance. The problem of personalized guidance has a long history in the area of adaptive educational systems (AES). In fact, the very first recognized AES SCHOLAR (Carbonell, 1970) focused on guiding students to the most relevant facts and questions about the geography of South America. The SCHOLAR functionality was based on a domain model in the form of a semantic network and an overlay student model. Since that time, a considerable share of research in the field of AES has focused on different kinds of personalized guidance, and the majority of this work relied heavily on domain modeling—which makes these two research directions heavily interconnected.

The purpose of this technical report is to provide an exemplar for conducting an analysis of a qualitative dataset using machine learning techniques. Qualitative data are measured or expressed as a natural language description (e.g., category or attribute) rather than numbers as in quantitative datasets. It is often difficult to evaluate the relationship between qualitative variables of interest and outcomes (dependent variables), but machine learning techniques offer simplified methods to classify these outcomes. WEKA, the Waikato Environment for Knowledge Analysis, is a popular set of machine learning algorithms developed at the University of Waikato in New Zealand, which can be used to analyze both qualitative and quantitative data. To illustrate the use of WEKA on a qualitative dataset, we selected a known set of primate species with the desire to classify them into 1 of 3 classes (prosimians, monkeys, and apes) based on 7 qualitative attributes. When an existing dataset is used with known relationships, it allows us to evaluate a large number of WEKA algorithms in a relatively short time and validate their accuracy with the goal of identifying best practices for analyzing qualitative data.

The tutorial described in this report was conducted at the Human-Computer Interaction International Augmented Cognition Conference in Creta Maris, Crete, Greece, in June 2014. The purpose of this tutorial was 3-fold: 1) familiarize participants with the fundamental principles of adaptive Intelligent Tutoring System (ITS) design; 2) illustrate how ITS design influences/enables self-regulated learning; and 3) discuss the need for standards for authoring of ITS, modeling of learners and experts, automated instructional strategies, and methods of analysis for ITS technologies., The original document contains color images.

Individual differences in the Need for Cognition (NFC) have been demonstrated to impact academic achievement, problem solving, and many classic psychology effects. The Self-Reference Effect (SRE) has been consistently shown to assist in recall and has been applied in education through context personalization of learning material. The current study examined if the SRE could aid learning for an applied deductive reasoning skill (completing a logic grid puzzle), and if an individual s NFC would impact learning outcomes. There were 134 participants that interacted with a computer-based tutoring system to teach them how to solve logic grid puzzles. In a logic grid puzzle, an individual uses deductive reasoning, and a set of clues to determine correct and incorrect information. During the tutorial, the puzzle and clues that the participants learned with either included their own name and the names of two close friends (self-reference), the names of three characters from the Harry Potter series, or names that were not expected to belong to the participants in the study. The names present in the tutorial interacted with NFC, so that those who were in the self-reference condition who were low NFC scored significantly lower on transfer performance than those that were high NFC. This suggests that individuals with high and low NFC may be impacted differently by self-reference, and it is important to consider its inclusion in instruction., The original document contains color images. Prepared in cooperation with Oak Ridge Associated Universities (ORAU), Oak Ridge, TN and University of Central Florida, Orlando, FL.

Scenario-based training exemplifies the learning-by-doing approach to human performance improvement. In this paper, we enumerate the advantages of incorporating automated scenario generation technologies into the traditional scenario development pipeline. An automated scenario generator is a system that creates training scenarios from scratch, augmenting human authoring to rapidly develop new scenarios, providing a richer diversity of tailored training opportunities, and delivering training scenarios on demand. We introduce a combinatorial optimization approach to scenario generation to deliver the requisite diversity and quality of scenarios while tailoring the scenarios to a particular learner's needs and abilities. We propose a set of evaluation metrics appropriate to scenario generation technologies and present preliminary evidence for the suitability of our approach compared to other scenario generation approaches., To be presented at the 7th International Conference on the Foundations of Digital Games, Raleigh, North Carolina, 30th May - 1st June, 2012. The original document contains color images.

In reflecting on a recent emphasis on self-directed learning using game-based simulations and virtual worlds, the authors considered key challenges in transforming serious games and virtual worlds into adaptive training tools. This article reflects specifically on the challenges and potential of cognitive modeling techniques for game-based tutors and game integration with computer-based. Finally, as we integrate tutors and games, a rubric is offered to identify some of the salient characteristics of adaptive game-based, tutoring systems., Presented at the 15th International Conference Artificial Intelligence in Education, Auckland, New Zealand, June 28-July 1, 2011

A renaissance in the research and development of computer-based, tutoring systems over the last ten years is motivating scientists to ponder the application of intelligent tutors and coaches in more challenging team training problem spaces where human tutors are either unavailable or impractical. This paper reviews some of the challenges and emerging technologies (tools and methods) that might influence the development of adaptive, intelligent tutors for geographically-distributed team training. Team tutoring presents many challenges. Even human tutors struggle to develop team cohesion, coordinate roles and responsibilities of team members and assess their contributions. Computer-based tutors face additional challenges: sensing and assessing the cognitive state (including affect) of each team member in near real-time to understand each team member's readiness to learn (e.g., their engagement and motivation); measuring team performance; perceiving and weighing team member contributions to team performance; and selecting instructional strategies that will optimize team performance. Emerging sensing technologies are showing promise as enablers of computer-based perception of each team member's behavior and physiology with the goal of predicting unobserved variables (e.g., cognitive state). Along with performance measures, historical and self-reported data, behavioral and physiological measures can provide the tutor with the information needed to model the trainee's state and their relationship with other team members and the tutor. Accurate (and timely) trainee and team state information (e.g., performance, competency, trust) are considered to be determining factors for the team tutor to select appropriate instructional strategies (e.g., support, direction) for optimal team performance., Presented at the Interservice/Industry Training, Simulation, and Education Conference (I/ITSEC), Nov 28-Dec 1, 2011, Orlando, FL.

Over the past two decades, the use of agent-based technology within simulated training environments has increased. Intelligent Tutoring Systems (ITS) technology may include reactive or proactive simulation agents that monitor and support computer-based training without human tutors. Reactive agents are able to provide hints and feedback on trainee performance within static scenarios. Based on the trainee's competency and their progress toward training objectives, proactive ITS use computational methods in real-time to decide when to change content, complexity and/or instructional methods within a training scenario (Niehaus & Riedl, 2009). This paper evaluates the advantages and disadvantages of reactive and proactive agents in computer-based tutoring systems; and discusses design considerations for the use of reactive and proactive agents in training simulations. Historically, intelligent tutoring agents have been simple, passive observers within simulation environments. These reactive agents monitor the trainee's progress and provide hints or other feedback only when there is sufficient variance from expected norms. Reactive agent actions are often based on simple heuristics or scripted behaviors. This can be desirable if the goal of the training is repeatability. However, reactive agents often know little about the trainee and the training context beyond performance data. Proactive agents have a higher computational cost in that they need to sense and understand more about the trainee, environment and training context, but are better able to predict trainee needs and adapt both feedback and scenario content. Complex military scenarios (e.g. ill-defined domains like bilateral negotiations) provide the opportunity to use more proactive agent techniques in assessing individual and team performance, and in adapting training scenarios to maintain challenge and flow., To be presented at the Interservice/Industry Training, Simulation, and Education Conference (I/ITSEC), 28 Nov-1 Dec 2011, Orlando, FL.

For our purposes, machine perception is defined as the ability of a computer-based training system to sense the behavior and affective state (e.g. mood or emotions) of trainees and interpret whether they are engaged, bored, frustrated, confused or even hostile during the training process. This paper puts forward the notion that the maturation of machine perception of trainee affect is critically important to optimizing learning for individuals and teams in embedded virtual simulations and other isolated training environments. Embedded training applications within operational platforms (e.g. tanks, aircraft, ships and individual Warfighting systems) continue to be explored today in many NATO countries (e.g. United States, Germany and the Netherlands). The lack of human tutors within operational platforms limits the understanding of each trainee's affective state and the completeness of the trainee model, the representation of the trainee's state within intelligent tutoring systems. Tutor technology is currently not sufficiently mature to provide accurate, portable, affordable, passive and effective sensing and interpretation of the trainee's affective state and limits the adaptability and effectiveness of the instruction in today's embedded training systems. This paper rationalizes the need for machine perception of affect in future embedded virtual simulations., See also ADA562526. RTO-MP-HFM-169 Human Dimensions in Embedded Virtual Simulation (Les dimensions humaines dans la simulation virtuelle integree)., The original document contains color images.