Projects
Eliciting Adaptive Sequences for Online Learning (funded by NSERC)
Adaptive learning aims to provide efficient, effective, and customized learning paths to engage each student. The intent of adaptive learning systems is not only to use proficiency and determine what a student really knows but also to move students accurately and logically through a sequential learning path to attain the prescribed learning outcomes and skill mastery. These adaptive learning systems are quickly emerging but are still in the experimental stages. Although they have many benefits – such as providing greater time efficiency and focused remediation – it is challenging to develop them. Adaptively sequencing content and assessment is the core of an adaptive learning system. Existing adaptive sequencing approaches, such as rule-based approach, the partially observed Markov decision process (POMDP) based framework, are either ineffective or difficult to develop. Recent advances in learning analytics, data mining, and machine learning, in particular, multi-armed bandit algorithms, present new opportunities for building effective adaptive sequencing algorithms for online learning. The multi-armed bandit family of algorithms is named after the problem for a gambler who must decide which arm of a “multi-armed bandit” slot machine to pull to maximize the total reward in a series of trials. These algorithms are data-driven and can balance exploration and exploitation and make sequential decisions under uncertainty. Thus, they are particularly relevant to decision-making about alternative pedagogies and lend themselves quite naturally to the problem of determining adaptive sequences in online learning environments. The objectives of this research are (1) To design a systematic methodology that includes a suite of algorithms based on the multi-armed framework that will work together with a hierarchical pedagogical model and a student model to produce adaptive sequences of knowledge components, learning activities, and formative assessment questions in an adaptive learning system. (2) To use simulations of students and courses to explore the performance of the designed algorithms and to optimize metrics or reward strategies; and (3) To develop a prototype of adaptive learning systems that will enable evaluations of the performance of the designed algorithms in real online courses. The long-term goal of this research is to enable future online learning systems to provide adaptively altering learning sequences of content and activities in real time that will best fit the student’s needs and knowledge states. Such systems are expected to make student learning not only easier but also far more efficient.
Interoperability between ALS and OER (Funded by AU IDEA LAB)
Both adaptive learning (AL) and open educational resources (OER) are two different but related innovations for addressing some of the cost, quality, and access challenges that most institutions are facing. AL is about the “how” of teaching and learning and OER as the “what.” It has been showed that combining the two innovations works better by The American Women’s College at Bay Path University. It would allow us to tap the benefits of each while simultaneously realizing unanticipated advantages. One of the benefits of OER is that it can be used to better meet the resource requirements of adaptive learning which requires the inclusion of multiple resources for each concept. OER can provide alternative or supplemental learning materials. This project explores the interoperability between adaptive learning systems (ALS) and OER by using LTI 1.3.
COD
COD is a MAS-based decision support system for course-offering determination. We studied using student preference elicitation, preference reasoning, voting-based preference aggregation, group decision making, agent negotiation techniques of MAS to solve planning, and scheduling issues in program planning and course-offering determination for educational programs.
Jason in Wonderland
Integrating a Multi-Agent System with a Virtual World
Grant has successfully completed the MSc IS program under the supervision of Dr Lin with a final project focusing on using intelligent software agents (MAS) (Jason) to control non-player characters (NPCs) in Open Wonderland (http://openwonderland.org/), a 3D virtual world. The project was funded by NSERC discovery grant and CFI of Canada. The contribution of this work lies in the area of integrating these systems together and coming up with a solution that allows users to create more interesting NPCs and scenarios in Open Wonderland. We show that MAS is the right approach to realizing more believable immersive virtual worlds for e-learning, e-health, etc. and the work that We have done is a step towards realizing that. The demos can be viewed from YouTube (Part 1):
and Part 2 and Part 3.
Game-like QuizMASter in a Virtual World
QuizMASter is an educational game being developed by a group of researchers at Athabasca University. It helps students perform adaptive testing and collaborative learning through friendly competition. Conceptually, QuizMASter is designed to be similar to a TV game show, where a small group of contestants compete by answering questions presented by the game show host. The game follows simple rules, whereby a contestant gets a point for answering the question correctly. The contestant with the highest points wins. However, QuizMASter is different from a TV game show, in that, every participant gets his or her separate perspective view and experience of the game on three different levels:
(1) QuizMASter's host is created using evolver avatar and controlled by Multi-agent System artificial intelligence technologies such as (Jason, Cartago) along with open wonderland's NPC module. The facial features of QuizMASter's host is morphed to that of the contestant. Using OWL's security feature, OWL object's visibility can be customized so that it is visible to certain clients and invisible to others. We exploited this feature of open wonderland and created customized host for every contestant whereby, the host possesses 75% resemblance to contestant's facial features and has the same skin tone as the contestant.
(2) The movements of every contestant are observed and recorded by an intelligent agent. After every single round, if the contestant is looking away from the host, his or her look direction is changed, so that he/she is facing the host. Furthermore, the contestant is requested by the host to pay better attention during subsequent rounds. If however, the contestant is found to have maintained eye contact throughout a particular round, he/she is thanked by the host at the end of the round.
(3) The Avatar project of Open Wonderland offers many capabilities to an evolver avatar in order to deliver facial expressions and gestures such as clapping, public speaking, cheering etc. These capabilities were combined with MARYTTS (Text-to-speech Synthesis technology) and other sound effects to provide an interactive and engaging environment for the contestant. OWL's security feature of Object's visibility/invisibility was also used in order to provide the contestants with customized situations/or experience where by, the host NPC, along with the crowd NPC's are empathetic towards the contestant if he/she gets a particular question wrong. Conversely, if the contestant get's a point during a particular round, he/she experiences praises from the host and cheering and clapping gestures from the crowd.
The above three levels of contestant's perspectives are designed using Transformed Social Interaction theoretical principles of 1) Self-Representation 2) Sensory-Abilities and 3) Situational-Context respectively. The whole objective of QuizMASter is to use the above three theoretical principles in the quest to achieve a more engaging learning environment for students.
The short video demonstrates version 1 of QuizMASter is here.
We are enhancing the intelligence capabilities and animations (such as lip-synchronization) of the game.