The paper introduces LT‑MKT, a new method for multi‑domain knowledge tracing that incorporates cognitive load and knowledge transfer. It constructs a multi‑domain hierarchical graph using textual information from questions and concepts, then explicitly models cross‑domain temporal and knowledge features to capture cognitive load effects. A knowledge transfer module further captures propagation of knowledge states within and across domains, leading to more accurate predictions of students’ future performance.
arXiv:2603. 02830v2 Announce Type: replace-cross Abstract: Predicting future student responses to questions is particularly valuable for educational learning platforms where it enables effective interventions.
The paper introduces TIZSKT, a zero‑shot cross‑domain knowledge tracing model that separates behavioral sequence modeling from domain‑local historical evidence construction. It uses text as an index via a frozen text encoder and unsupervised per‑domain projection to build question addresses, then aggregates a student’s past outcomes into a question‑specific historical correctness summary. Experiments on three public datasets show that TIZSKT outperforms baselines in mean AUC and accuracy, with ablations confirming the importance of text addressing and cross‑segment continuation.
The paper introduces Process-aware Language Cognitive Diagnosis (PLCD), a framework that replaces traditional ID-based embeddings in Cognitive Diagnosis Models with language-derived structures and response records. PLCD employs large language models to build concept schemas and cognitive process graphs, and uses a Language-to-Cognition Mapper with DA-MoE experts and contrastive learning to map textual evidence into a unified cognitive space. Experiments demonstrate that PLCD outperforms conventional baselines in student performance prediction and shows strong cognitive transfer, improving cold-start robustness and cognitive grounding.
arXiv:2607. 13103v1 Announce Type: cross Abstract: Knowledge tracing (KT) aims to predict students' future performance by modeling their evolving knowledge states from historical interactions.
Temporal knowledge graphs are central to many uses of the Semantic Web, but existing completion methods assume the entities, relation names, and timestamps to be reasoned about are already known at training time, restricting each model to a single graph and vocabulary. We propose FITTER, the first fully-inductive structural model for temporal knowledge graph link prediction that supports cross-domain transfer: the inference graph may contain entirely unseen entities, relation names, and timestamps drawn from a different domain.
arXiv:2608. 10668v1 Announce Type: new Abstract: Temporal knowledge graphs are central to many uses of the Semantic Web, but existing completion methods assume the entities, relation names, and timestamps to be reasoned about are already known at training time, restricting each model to a single graph and vocabulary.
The paper investigates why curriculum learning—ordering training data from easy to hard—varies in effectiveness across reasoning tasks. By studying optimization dynamics, the authors introduce Relative Transfer, a measure of cross‑difficulty knowledge transfer, and use it to create Transfer‑aware Dynamic Curriculum Sampling (TDCS). Experiments show TDCS outperforms existing scheduling strategies on multiple reasoning benchmarks, offering a unified optimization‑based explanation for curriculum learning.
arXiv:2608. 11788v1 Announce Type: cross Abstract: Continual pre-training (CPT) has been widely adopted as a method for domain adaptation in large language models.
The paper investigates how language models acquire knowledge over the course of pretraining by using point-in-time models trained on web snapshots from different years. It introduces a temporal variability score (TVS) that measures how consistently a knowledge item is represented across these snapshots. The study finds that items with high TVS improve more rapidly later in training and outperform low‑TVS items in most benchmark settings, especially on knowledge‑intensive tasks.
The paper explores using semantic text embeddings to automatically build educational taxonomies by clustering math question embeddings. It compares these embedding‑based taxonomies to existing Common Core State Standards (CCSS) taxonomies in two ASSISTments datasets, finding substantial but incomplete overlap. When applied to Deep Knowledge Tracing, the new taxonomies improve predictive performance by about 3‑5 AUC percentage points.
The paper investigates three fusion paradigms—Merge, Mix RL, and multi‑teacher on‑policy distillation (MOPD)—for consolidating reinforcement learning with verifiable rewards (RLVR) across multiple domains. Experiments across model scales and a multi‑domain benchmark show that while overall performance differences are small, significant gaps can appear on specific tasks, and each method exhibits distinct training dynamics and constraints. Practical guidelines are offered: Merge for cheap fusion when experts exist, Mix RL for unified training with adjustable domain mixtures, and MOPD when preserving domain‑specific gains is paramount.
arXiv:2508. 17092v2 Announce Type: replace-cross Abstract: Knowledge Tracing (KT) aims to predict a student's future performance based on their sequence of interactions with learning content.