Research map / Video / Video segmentation and tracking
MRMLREC: A Two-Stage Approach for Addressing Data Sparsity in MOOC Video Recommendation (Student Abstract)
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes MRMLREC, a two-stage MOOC video recommender that builds temporal, course, and knowledge-concept relational graphs encoded by GCN/GAT and an LSTM, then predicts course, concept, and video levels with beam search to address data sparsity.
Problem
Interactive data sparsity in MOOC video recommendation, where learners' sparse interaction data makes predicting the next learning video difficult and short-term learning needs are ignored.
Approach
Combines multi-relational graph representations (temporal sequence, course-video, knowledge-concept-video) fused by a variant LSTM with multi-level prediction (course, concept, video) pruned by beam search, plus an optional personalization/diversification layer.
Key results
- On the Computer dataset, MRMLREC reaches HR@10 0.7887 and NDCG@10 0.7261, versus 0.4950 / 0.2219 for the best baseline IntGNN.
- On MOOCCube, MRMLREC reaches HR@10 0.7932 and NDCG@10 0.7440, versus 0.1426 / 0.0724 for the best baseline TiCoSeRec.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use MRMLREC instead of IntGNN | MOOC video recommendation · Computer | HR@10: 0.4950 -> 0.7887 | Helped |
| Use MRMLREC instead of IntGNN | MOOC video recommendation · Computer | NDCG@10: 0.2219 -> 0.7261 | Helped |
| Use MRMLREC instead of IntGNN | MOOC video recommendation · Computer | MRR: 0.1933 -> 0.7075 | Helped |
| Use MRMLREC instead of TiCoSeRec | MOOC video recommendation · MOOCCube | HR@10: 0.1426 -> 0.7932 | Helped |
| Use MRMLREC instead of TiCoSeRec | MOOC video recommendation · MOOCCube | NDCG@10: 0.0724 -> 0.7440 | Helped |
| Use MRMLREC instead of TiCoSeRec | MOOC video recommendation · MOOCCube | MRR: 0.0571 -> 0.7294 | Helped |
Limitations
- Evaluation limited to two datasets from one platform (Xuetang X) and four baselines.
- Abstract is truncated; the optional personalization/diversification layer is described but not evaluated separately in the text shown.
Experimental setup
Datasets: Computer MOOCCube
Benchmarks: Computer MOOCCube
Metrics: HR@10 NDCG@10 MRR
Baselines: COTREC lightGCN TiCoSeRec IntGNN
Builds on: LSTM GCN GAT beam search
Closest papers by meaning
- Multi-type MOOCs Recommendation: Leveraging Deep Multi-Relational Representation and Hierarchical Reasoning (AAAI 2025 · similarity 0.91)
- LGMRec: Local and Global Graph Learning for Multimodal Recommendation (AAAI 2024 · similarity 0.84)
- TV-Rec: Time-Variant Convolutional Filter for Sequential Recommendation (NeurIPS 2025 · similarity 0.83)
- ReSpec: Relevance and Specificity Grounded Online Filtering for Learning on Video-Text Data Streams (CVPR 2025 · similarity 0.83)
- Future Sight and Tough Fights: Revolutionizing Sequential Recommendation with FENRec (AAAI 2025 · similarity 0.83)
- Augmenting Sequential Recommendation with Balanced Relevance and Diversity (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.