Text indexing improves zero-shot knowledge tracing across platforms

Text as an Index for Zero-Shot Cross-Domain Knowledge Tracing

Computation and LanguageComputers and SocietyMachine Learning

Summary

Predicting how students will answer questions on new learning platforms is hard because there is no past data to learn from. The authors propose a method called TIZSKT that uses question text as an index to look up a student's past answers to similar questions without relying on direct matching of question meanings. This lets the system better predict student performance on new platforms without retraining. Tests on multiple datasets showed TIZSKT outperformed other methods, especially by keeping longer student history and using text to relate questions.

What this means in practice

  • For education technology developers: Improve adaptive learning systems on new educational platforms without existing student response data by indexing question texts for personalized predictions.
  • For online course platform engineers: Deploy knowledge tracing that predicts student answers early when starting a new course or platform using public question texts and limited interaction history.

Authors

Yin Yi, Haiping Ma, Siyu Song, Ziwen Wang, Shangshang Yang

Abstract

Knowledge tracing (KT) predicts a student's next response from their history. New platforms often have question texts but no response logs for training, motivating zero-shot cross-domain KT. Existing transfer and cold-start methods use question semantics, but cross-platform semantic discrepancies limit transferability. We propose TIZSKT, which separates behavioral sequence modeling from domain-local historical evidence construction and uses text as an index rather than a directly transferable predictive representation. The Mastery-Dynamics Encoder learns correctness dynamics without question identifiers or text embeddings. Domain-Local Question Addressing builds fixed question addresses using a frozen text encoder and an unsupervised per-domain projection. The Text-Addressing Kernel uses these addresses to aggregate the current student's previous outcomes into a question-specific historical correctness summary. Cross-Segment Continuation retains access to earlier responses without extending the behavioral encoder's input sequence. Evidence-Volume Read-Out preserves the total kernel mass discarded by normalization and combines its log-compressed value with the behavioral state and historical correctness summary to predict the next response. All trainable parameters are learned from source domains; target inference uses public question texts and the current student's observed history without target-domain training. Leave-one-domain-out experiments on three public datasets show that TIZSKT achieves the highest mean AUC and accuracy among evaluated baselines on all target domains. Ablations confirm the contributions of text addressing and cross-segment continuation.