Skip to main content Skip to navigation

Deep Knowledge Tracing with Side Information

Project Overview

The document explores the application of generative AI in education, focusing on a novel framework known as Deep Knowledge Tracing with Side Information (DKTS), which enhances knowledge tracing within intelligent tutoring systems. By integrating sequential dependencies and intrinsic relationships among questions, DKTS aims to provide more accurate predictions of student performance and knowledge states. The framework employs advanced deep learning techniques and has demonstrated superior effectiveness on real educational datasets, showcasing its potential to personalize learning experiences. This innovative approach not only addresses the challenges of traditional tutoring systems but also highlights the transformative role that generative AI can play in tailoring educational interventions to meet individual student needs, thereby fostering improved educational outcomes. Overall, the findings underscore the significance of leveraging generative AI technologies to create more adaptive and responsive learning environments in education.

Key Applications

Deep Knowledge Tracing with Side Information (DKTS)

Context: Intelligent tutoring systems for personalized education, targeting students using online platforms.

Implementation: Utilizes deep neural networks to capture both sequential dependencies and intrinsic question relations from student response data.

Outcomes: Improved prediction accuracy of student performance on future questions, outperforming traditional models.

Challenges: Incorporating complex question relations can be computationally intensive and may require large datasets for effective training.

Implementation Barriers

Technical Barrier

The complexity of human learning processes makes knowledge tracing inherently challenging.

Proposed Solutions: Leveraging large datasets and recent advances in machine learning technologies to enhance model accuracy.

Data Barrier

High-quality and relevant educational data is crucial for training effective models.

Proposed Solutions: Collecting and cleaning data from popular educational platforms to ensure quality input for models.

Project Team

Zhiwei Wang

Researcher

Xiaoqin Feng

Researcher

Jiliang Tang

Researcher

Gale Yan Huang

Researcher

Zitao Liu

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Zhiwei Wang, Xiaoqin Feng, Jiliang Tang, Gale Yan Huang, Zitao Liu

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies