Knowledge Is Power: Harnessing Large Language Models for Enhanced Cognitive Diagnosis

Authors: Zhiang Dong, Jingyuan Chen, Fei Wu

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on several real-world datasets demonstrate the effectiveness of our proposed framework. Our code and datasets are available at https://github.com/ Player Dza/KCD.
Researcher Affiliation Academia Zhejiang University EMAIL
Pseudocode No The paper describes the methodology in detail, including framework overview, LLM diagnosis, and cognitive level alignment, but does not present a clearly labeled pseudocode or algorithm block.
Open Source Code Yes Our code and datasets are available at https://github.com/ Player Dza/KCD.
Open Datasets Yes Experiments on several public datasets with different CDMs demonstrate the effectiveness of our framework. Our code and datasets are available at https://github.com/ Player Dza/KCD.
Dataset Splits Yes The datasets are divided into training, validation, and testing sets, with a ratio of 8:1:1.
Hardware Specification Yes All the experiments are conducted on a Ge Force RTX 3090 GPU.
Software Dependencies Yes We utilize Py Torch to implement both the baseline methods and our proposed KCD framework. For the baseline models, We use the default hyper-parameters as stated in their papers and for KCD, we use the same hyper-parameter settings, such as training epoch, learning rate, and batch size. We employ Chat GPT to represent LLMs (specifically, gpt-3.5-turbo-16k) and textembedding-ada002 as the text embedding model.
Experiment Setup Yes We use the default hyper-parameters as stated in their papers and for KCD, we use the same hyper-parameter settings, such as training epoch, learning rate, and batch size. We employ Chat GPT to represent LLMs (specifically, gpt-3.5-turbo-16k) and textembedding-ada002 as the text embedding model. All the experiments are conducted on a Ge Force RTX 3090 GPU. We train the model on train set and at the end of each epoch, we evaluate the model on the validation set. The hyperparameter α, β, and λ was set to 0.04, 0.015, and 0.2.