AAAI2023 Global Knowledge Tracing Challenge

17 Oct 2022 By

Challenge Website

The Association for the Advancement of Artificial Intelligence (AAAI) is a prestigious organization in the field of artificial intelligence and computer science. It was founded by Allen Newell, Marvin Minsky, John McCarthy, and other luminaries in the field. As a leading organization, AAAI has become a platform for scholars and professionals to showcase their cutting-edge research in AI and explore its academic applications.

In February 2023, the AAAI2023 Smart Education Symposium will be held in conjunction with the AAAI 2023 conference in Washington, DC. The AAAI2023 Global Knowledge Tracing Challenge, initiated by the Guangdong Smart Education Research Institute of Jinan University, in collaboration with several well-known educational technology companies and universities, is one of the highlights of the symposium.

The Challenge invites researchers and practitioners worldwide to investigate the opportunities to improve student assessment performance through knowledge tracing approaches with rich side information. Knowledge tracing involves using students’ historical learning interaction data to model their knowledge mastery over time and predict their future interaction performance. The challenge provides a large-scale student assessment dataset with rich textual and structural information, including 18,066 students, 7,652 questions, and 5 million interactions. Each question includes text, answer explanations, and corresponding knowledge points.

The AAAI2023 Global Knowledge Tracing Challenge aims to develop new techniques and algorithms that leverage this rich side information to better predict students’ abilities and improve their learning outcomes. The competition is built on pyKT, a python library based on PyTorch that simplifies the process of training deep learning-based knowledge tracing models. The library has many advantages, such as being easy to use and having all of its modules carefully designed and developed. It is beginner-friendly, and running models only requires three steps, making it easier to build new models and create new datasets. Moreover, pyKT is designed for research and includes a review process for all models and datasets to ensure that the code is correct, and all details, such as data processing, data splitting, different prediction models, and model parameter optimization, are open-sourced.

The competition attracted 37 different organizations from eight countries, including the United States, the United Kingdom, Australia, and Singapore, with a total of 116 participants. Well-known companies and research institutes, such as Ant Group, NetEase, A*STAR, and TagHive, participated, as well as world-renowned universities, such as Carnegie Mellon University, Columbia University, Cornell University, Massachusetts Institute of Technology, Korea University, and the University of Edinburgh etc.

The total awards for the AAAI2023 Global Knowledge Tracing Challenge are $1900, with the final submission deadline set for December 31, 2022. By participating in the challenge, researchers and practitioners have the opportunity to contribute to the advancement of the education system, predict students’ abilities more accurately, and help students learn better and faster. Furthermore, the challenge is an excellent platform for researchers and practitioners to collaborate, share their ideas, and explore the frontiers of knowledge tracing techniques.