Cross-Platform Domain Adaptation for Multi-Modal MOOC Learner Satisfaction Prediction

TL;DR

This paper introduces ADAPT-MS, a novel framework for cross-platform MOOC learner satisfaction prediction that effectively adapts to new platforms with limited or no labeled data, improving accuracy over existing methods.

Contribution

The paper proposes ADAPT-MS, a comprehensive platform-adaptive model combining text encoding, domain alignment, rating bias correction, and modality handling for cross-platform satisfaction prediction.

Findings

ADAPT-MS achieves RMSE of 0.66 in unsupervised adaptation.

With 1000 labeled samples, RMSE improves to 0.60.

The framework outperforms naive pooling, domain alignment without calibration, and full fine-tuning.

Abstract

Learner satisfaction prediction from MOOC reviews and behavioral logs is valuable for course quality improvement and platform operations. In practice, models trained on one platform degrade significantly when deployed on another due to domain shift in review style, learner population, behavioral logging schemas, and platform-specific rating norms. We study \textbf{cross-platform domain adaptation} for multi-modal MOOC satisfaction prediction under limited or absent target-platform labels. We propose \textbf{ADAPT-MS}, a platform-adaptive framework that (i) encodes review text with a frozen LLM encoder and behavioral traces with a canonical-vocabulary MLP, (ii) aligns cross-platform representations via domain-adversarial training with gradient reversal, (iii) corrects platform-specific rating bias through a latent-variable calibration layer, and (iv) handles missing behavioral modalities via gated fusion with modality dropout. Experiments on a multi-platform MOOC dataset spanning three major platforms demonstrate that ADAPT-MS achieves target-platform RMSE of 0.66 in the unsupervised setting (zero labeled target samples) and 0.60 with 1000 labeled target samples, outperforming strong baselines including naive pooling, domain-adversarial alignment without calibration, and full fine-tuning. Ablation studies confirm the independent contribution of each component, and few-shot adaptation curves demonstrate stable improvement even with as few as 50 labeled target samples.