Cognitive Decoupling and Adaptive Fusion for Knowledge Tracing

We added cdfkt into our pyKT package.

The link is here and the API is here.

Original paper can be found at Wu et al. “Reducing Causal Attribution Ambiguity in Knowledge Tracing via Cognitive Decoupling and Adaptive Fusion.” Information Fusion (2026): 104648.

cdfkt reduces causal attribution ambiguity in knowledge tracing via cognitive decoupling and adaptive fusion. It decouples student cognitive state into situational adaptability (modeled by an exponential-decay kernel for short-term fluctuations) and long-range knowledge accumulation (modeled by a Mamba state-space model), then fuses them through an adaptive gate for response prediction.