Current Projects

• I am the principal investigator of NSERC-funded project entitled “Eliciting Adaptive Sequences for Online Learning” (2021-2026). 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.


• I am the Institutional Lead of Athabasca University for Space and Defence Technology Alberta Project -- SDTech_AB (2024-2028).

Intelligent Educational Systems Research Group

Research Outcomes

• Adaptive QuizMaster: Student Login | Instructor Login
  Project Description: Formative assessments help teachers identify concepts that learners are struggling to understand, skills they are having difficulty acquiring, or learning standards they have not yet achieved so that adjustments can be made to lesson contents, instructional approaches, and academic support. Formative assessment should be viewed as an integral part of instruction, and it should be used in real time for guiding the learning process. To identify the gaps quickly and accurately between what is known and what is aimed to be known, adaptive testing has been proposed to replace static assessments. However, existing latent factor based adaptive assessment methods need complex question pool calibration and use the debatable premise that different questions measure one common latent trait. Also, the knowledge tracing approaches are hard to implement and use. To overcome these limitations, we have been developing and testing an online adaptive formative assessment system, Adaptive Quizmaster, for learning online courses. It is an LTI-enabled system that can be integrated with other learning platforms. The objective of this research project is to develop and test new question selection mechanisms using Reinforcement Learning (RL). With this framework, we aim to create formative assessments that can adaptively select questions from a question bank according to students’ responses with as few questions as possible while ensuring the accuracy of assessments.
  
  
• QuizGen: Experimental System Login:
  Project Description: In the rapidly evolving landscape of educational technology, the potential of large language models (LLMs) like ChatGPT, Midjourney, MS Copilot, Claude, and open-source AI Llama, and is becoming increasingly apparent. These advanced AI tools typically utilize a chat-based interface that simulates a one-on-one conversation. However, when it comes to generating complex educational materials, e.g. question generation, these interfaces often fall short. The inherent challenge lies in the difficulty of articulating specific requirements succinctly, achieving clarity and alignment often necessitates extended dialogue.
  This NSERC USRA research project aims to bridge this gap by
  • assessing the state-of-the-art of LLM-based AI tools in question generation,
  • overcoming the current limitations by using Retrieval-Augmented Generation (RAG) and new prompting methods, to mitigate the issue of AI hallucinations, where the model generates plausible but incorrect or nonsensical information
  • developing a web-based LLM-powered AI system designed for human-AI interaction, specifically dedicating to the development of automated AI assistant capable of executing actions via a User Interface (UI), targeting the generation of formative assessment questions. The UI of this innovative system is crafted to reflect a real-life collaborative approach between teachers and AI, moving beyond traditional chat-based models. We have tested this interface against a set of quality validation criteria (QVC) grounded in instructional design principles to ensure its effectiveness and usability.
  


• Algorithm Visulization and Game-based Learning to learn Game-theory (2016-present)

by Manuel Gentes, Gurpreet Singh, Nitin Dogra, Diptanshu Mandal, and Dr Oscar Lin (thanks to the support of MITACS and AU)
Bubble Sorting | Selection Sorting | Insertion Sorting | Merge Sorting | Quick Sorting
Centipede Game | El Farol Bar Problem | Rock, Paper, and Scissors Game


• eCourseGuard Collaborative 3D User Interfaces for Virtual Classrooms in Distance Education (2007-2009)
  It is a Special Project of Athabasca University, and Sun Campus Ambassador at AU. There are many 3D User Interfaces (UI) application areas, such as collaborative product design, virtual classrooms for distance education, virtual training, etc, in which collaborative, multi-user work is the norm. But a collaborative 3D UI for teamwork is non-trivial. It is not simply the sum of several single-user UIs. Why 3D for Collaboration? The 3D space provides a natural way to organize multiple, simultaneous conversations. Likewise, the arrangement of the objects within the space provides conversational context. If other avatars are gathering near the entrance to a virtual conference room, it is a good guess that they are about to attend a meeting in that space. It is then natural to talk to those people about the content or timing of the meeting, just as you would if attending a physical meeting. In terms of data sharing, looking at objects together is a natural activity. With the 3D spatial cues, each person can get an immediate sense of what the other collaborators can and cannot. To facilitate collaboration, we need to face a multitude of interface challenges, such as:
  
  1. awareness, (Who is here? Where are they? What are they doing?),
  2. communication (speech, pointing, facial expressions), and
  3. "floor control" (who is working on that object? How do I get the object?).


• Game-Like QuizMASter (2009-2014)
Jason in Wonderland ---- Integrating a Multi-Agent System with a Virtual World
We (mainly by MSc IS graduates Grant McClure, Jeanne Blair) successfully completed a project on using Multiagent Systems (MAS) (Jason) to control non-player characters (NPCs) in Open Wonderland, 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 1 , 2 , 3 and Game-like QuizMASter, which is an experimental system for educational games developed by a group of researchers at Intelligent Educational Systems research group at Athabasca University of Canada. It helps students perform adaptive testing and collaborative learning through friendly competition. Conceptually, Game-like 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.


• e-Advisor and COD (2005-2012)

e-Advisor is an intelligent system designed to assist students in open, online, and distance learning environments like Athabasca University with different background and job objectives to plan their study and select appropriate online courses. The system can be incorporated into online learning systems like Moodle to enhance their functionality, while minimizing the use of scarce program advising time. We propose a loosely-coupled architecture in which Web services are used for integrating various distributed information resources and supporting intelligent software agents in the processes of decision-making. Using Semantic Web technologies, the learner information Web service relies on a learner ontology based on the IMS LIP specifications while the program information Web Service uses domain ontologies and Petri nets formalism. The experimental version of the system has been in trial in 2005 with promising results.
COD is a MAS-based decision support system for course-offering determination. We studied using student preference elicitation, preference reasoning, voting-based preference ggregation, group decision making, agent negotiation techniques of MAS to solve planning, and scheduling issues in program planning and course-offering determination for educational programs.


• eCourseGuard Oil and Gas Well Scheduling System (2008-2011)
  IS a prototype of an intelligent resource allocation system, The project was funded by NSERC Engage. We collaborated with Encana Services Company Ltd. (Encana) and applied multiagent coordination mechanisms technologies, contract-net protocol (CNP), market-inspired coordination mechanism, auction-based resource allocation. Each agent uses bid prediction machine learning, use multi-armed bandit algorithms to solve the well scheduling and re-scheduling interval optimization problem in oil and gas production.


• eCourseGuard MARC to IEEE LOM CrossWalk (2002-2006)
  E-learning has the potential to provide the flexibility and wider access that is required for lifelong learning, but creating the digital resources needed for online course delivery requires a considerable investment and substantial effort. There are some pre-existing learning objects available for reuse in the design of educational events. However, supplying the metadata for each standard becomes repetitious, time-consuming, and tedious because of the diversity of learning objects, and the continuing growth in the number, size, and complexity of the content metadata standards. In order to minimize the amount of time needed to create and maintain the metadata and to maximize its usefulness to the widest possible community of users, there is a demand for developing crosswalks between different metadata standards. This project developed a crosswalk that converts MARC (MAchine-Readable Cataloging) metadata to IEEE LOM (Learning Object Metadata) and investigates the issues involved in the crosswalk development.


• eCourseGuard -- a Broken Link Checking Agent (2002-2005)
  The main goal of this project is to teat the feasibility of the architecture proposed. In Web-based education systems for distance and flexible learning, content for learning delivered to students is the core of the system. It is important to deliver current, correct, and complete course materials to students. Given the past ten years of experience with Web-based course delivering, the course material checking and update notification increased much workload of faculty and administrators. Therefore, we designed and developed an eCourseGuard. The eCourseGuard is an agent system that automates the course material checking and student notification. The eCourseGuard thoroughly and continually detects the link availability and content changes in the courses and swiftly notifies the instructors or students by email so that they can respond accordingly.