Skip to main content Skip to navigation

Predicting Knowledge Gain for MOOC Video Consumption

Project Overview

The document examines the role of generative AI in education, particularly through a research study aimed at predicting knowledge gain from Massive Open Online Course (MOOC) videos by utilizing multimodal features, such as slide visuals and spoken content. It highlights the automated assessment of learning resources, focusing on how various content features can forecast learning success. Key findings reveal that both multimedia elements and textual information significantly enhance the accuracy of knowledge gain predictions. The research identifies critical feature categories that contribute to this predictive capability, suggesting that future studies should leverage larger datasets and integrate user feedback to further refine prediction models. Overall, the document underscores the potential of generative AI to transform educational outcomes by improving assessments and personalized learning experiences.

Key Applications

Knowledge gain prediction for MOOC videos using multimodal features

Context: Online learning environments, targeting learners using MOOC platforms like edX

Implementation: The study employed machine learning classifiers to analyze video content and predict knowledge gain based on extracted features from slides and speech.

Outcomes: The research resulted in improved accuracy for predicting knowledge gain, highlighting the effectiveness of combining multimodal features.

Challenges: The study faced challenges related to feature selection and the need for a larger dataset to validate the findings.

Implementation Barriers

Data Limitations

Limited dataset size with only 22 unique feature samples, which hampers the robustness of the models. Future work should involve larger studies with more users and videos to replicate the results.

Proposed Solutions: Future work should involve larger studies with more users and videos to replicate the results.

Feature Complexity

Complexity of feature extraction and the need for a comprehensive analysis of various modalities. Utilizing automated tools and frameworks for feature extraction and analysis can streamline the process.

Proposed Solutions: Utilizing automated tools and frameworks for feature extraction and analysis to streamline the process.

Project Team

Christian Otto

Researcher

Markos Stamatakis

Researcher

Anett Hoppe

Researcher

Ralph Ewerth

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Christian Otto, Markos Stamatakis, Anett Hoppe, Ralph Ewerth

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