OATutor: An Open-source Adaptive Tutoring System and Curated Content Library for Learning Sciences Research

Title of the Paper

OATutor: An Open-source Adaptive Tutoring System and Curated Content Library for Learning Sciences Research

Paper Information

• Subject Area: Research on Adaptive Learning Systems and Educational Technology
• Keywords: Adaptive Learning, Intelligent Tutoring System, Open Source, Open Educational Resources, Content Creation, Design-driven Research, Reproducible Research

Research Background and Problem

• What problems or challenges did the authors identify?

  • Although Intelligent Tutoring Systems (ITS) have been proven to enhance learning outcomes, existing systems are proprietary and closed-source, which significantly limits researchers' ability to conduct replication studies or extend their designs.
  • There is a lack of an open, adaptive tutoring platform available for the research community, hindering the integration of Open Educational Resources (OER) with intelligent tutoring systems.

• Why is this problem important?

  • Open-source adaptive tutoring systems can save development costs for educational researchers and improve content creation efficiency. Additionally, they can promote technological openness, data sharing, and reproducible research in the field of education.

• Research Motivation and Related Work

  • Existing movements in open educational technology (e.g., open educational resources, open courses, open datasets, and algorithms) have significantly advanced educational science research.
  • The authors aim to bridge the gap between open educational resources and intelligent tutoring systems by developing an open-source adaptive tutoring system to lower research barriers and foster experimentation and innovation.

Solution

• What methods or solutions did the authors propose?

  • The authors proposed and developed an open-source adaptive tutoring system called OATutor (Open Adaptive Tutor).
  • OATutor includes a tutoring platform based on ITS principles, equipped with three algebra problem sets licensed under Creative Commons BY 4.0, and features knowledge tracking, an A/B testing framework, and LTI integration.
  • They created learner-facing authoring tools to facilitate the generation of adaptive content and invited the community to contribute content.

• What is innovative about this solution?

  • For the first time, a complete intelligent tutoring system has been open-sourced, covering content management, skill mastery estimation, adaptive problem selection, and integration with Learning Management Systems (LMS).
  • It supports a flexible experimental framework based on knowledge tracking and A/B testing, enabling exploration of educational research questions.
  • The content library is built collaboratively with learners, reducing the time and cost of content creation.

• What are the implementation steps? What key technologies were used?

  • Implementation Steps:
    1. Define the content structure and skill model architecture to support modular content organization.
    2. Design a problem selection algorithm based on knowledge tracking to estimate skill mastery.
    3. Create scalable content authoring tools using Google Spreadsheet and JSON formats for creation and conversion.
    4. Integrate with Learning Management Systems (LMS) and support user behavior data logging.
    5. Collect feedback during trials in community college classrooms to optimize system functionality.
  • Key Technologies:
    • Bayesian Knowledge Tracing (BKT) algorithm
    • React framework for front-end development
    • Firebase database for data storage
    • Python-based pyBKT library for training knowledge tracking models

Research Outcomes

• What specific outcomes were achieved?

  • Successfully built a complete and open-source adaptive tutoring system.
  • The initial system content includes three comprehensive algebra textbooks, available for direct use or extension by educational researchers.
  • Improved functionality through real classroom trials and multiple iterations, enabling teachers and researchers to run their own experimental frameworks.

• What advantages does it have compared to existing solutions?

  • The open-source nature lowers technical barriers for development and experimentation.
  • Allows researchers to customize experiments on knowledge tracking algorithms and tutoring processes.
  • Significantly improves content creation efficiency, with an average of only 11 minutes required to create a single problem.

• What were the experimental or evaluation results?

  • The system collected over 70,000 problem submission data points.
  • In community college trials, 150 students participated, and feedback indicated strong applicability for knowledge expansion purposes.
  • Learners provided over 2,000 pieces of feedback, significantly improving the system experience.

• Limitations and Future Directions

  • Limitations:
    • The system currently supports only algebra courses, with no content for other subjects.
    • It does not yet meet teachers' needs for full control over learning content (e.g., customizing or editing problems).
    • Supported problem types are limited (e.g., graphical problems are not yet implemented).
  • Future Directions:
    • Expand to other subjects and fields (e.g., statistics, computer science).
    • Develop more advanced machine-generated content tools to reduce the burden of manual creation.
    • Explore flexible design methods for granting teachers control over system content editing.
    • Evaluate the A/B testing effectiveness of OATutor compared to existing tutoring systems.