Using Knowledge Concept Aggregation towards Accurate Cognitive Diagnosis

Wang, Xinping; Huang, Caidie; Cai, Jinfang; Chen, Liang‐Yu

doi:10.1145/3459637.3482311

Cited by 13 publications

(9 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Three famous metrics, i.e., the Root Mean Square Error (RMSE) (Tian et al 2022), the Prediction Accuracy (ACC) (Tian et al 2021) and Area Under an ROC Curve (AUC) (Bradley 1997) were chosen to evaluate predictive performance. (Lord 1980), MIRT (Reckase 2009), NCD (Wang et al 2020a), CDGK (Wang et al 2021), ECD (Zhou et al 2021) and RCD (Gao et al 2021).…”

Section: Methodsmentioning

confidence: 99%

Enhancing Cognitive Diagnosis Using Un-interacted Exercises: A Collaboration-Aware Mixed Sampling Approach

Ma,

Wang,

Zhu

et al. 2024

AAAI

View full text Add to dashboard Cite

Cognitive diagnosis is a crucial task in computer-aided education, aimed at evaluating students' proficiency levels across various knowledge concepts through exercises. Current models, however, primarily rely on students' answered exercises, neglecting the complex and rich information contained in un-interacted exercises. While recent research has attempted to leverage the data within un-interacted exercises linked to interacted knowledge concepts, aiming to address the long-tail issue, these studies fail to fully explore the informative, un-interacted exercises related to broader knowledge concepts. This oversight results in diminished performance when these models are applied to comprehensive datasets. In response to this gap, we present the Collaborative-aware Mixed Exercise Sampling (CMES) framework, which can effectively exploit the information present in un-interacted exercises linked to un-interacted knowledge concepts. Specifically, we introduce a novel universal sampling module where the training samples comprise not merely raw data slices, but enhanced samples generated by combining weight-enhanced attention mixture techniques. Given the necessity of real response labels in cognitive diagnosis, we also propose a ranking-based pseudo feedback module to regulate students' responses on generated exercises. The versatility of the CMES framework bolsters existing models and improves their adaptability. Finally, we demonstrate the effectiveness and interpretability of our framework through comprehensive experiments on real-world datasets.

show abstract

Section: Methodsmentioning

confidence: 99%

Enhancing Cognitive Diagnosis Using Un-interacted Exercises: A Collaboration-Aware Mixed Sampling Approach

Ma,

Wang,

Zhu

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…The connection-driven cognitive diagnostic RCD [11] models the prior structural relationship between students' exercise interaction and knowledge points using a relation graph with multiple layers. CDGK [12] transforms knowledge concepts into a graph and only considers the leaf nodes of the knowledge concept tree to collect knowledge concepts and decrease the dimensions of the model.…”

Section: Related Workmentioning

confidence: 99%

Cognitive Diagnosis for Programming Domains

Huang,

Wang

2023

Atlantis Highlights in Computer Sciences

View full text Add to dashboard Cite

Cognitive diagnosis plays an important role in intelligent educational scenarios as a method that can reveal students' knowledge mastery. Existing cognitive diagnostic methods are mainly applicable to objective questions in core subject areas (e.g., mathematics); however, in the field of programming, where the questions are subjective, no research has been conducted to apply cognitive diagnostics to analyze and assess the impact of students' subjective answers on students' knowledge acquisition. Therefore, we explored how to design a cognitive diagnostic approach that can be applied in the programming domain. In this paper, we design a neural network-based cognitive diagnostic model, PECDM, where our approach not only exploits the interactions between the student factor and the exercise factor, but also includes the student answer source code and the difficulty of the exercise among the diagnostic factors, further considering that the student's knowledge mastery can be captured from the student's subjective answers (source code), and finally uses multiple fully connected layers to interactions are modeled, resulting in more accurate and interpretable diagnostics. We conduct relevant experiments on the codeforces dataset, and the experimental results show that the accuracy of our designed PECDM model is about 95%, which is better in terms of accuracy, rationality, and interpretability when compared with other cognitive diagnostic models.

show abstract

“…Therefore, DNN-based methods redesign the model structure by applying the deep neural network. The typical methods include NCDM [Wang et al 2020], which incorporates DNN to learn the exercising interactions, and CDGK [Wang et al 2021], which considers the knowledge concept aggregation to extend the NCDM. However, DNN-based methods do not consider the impact of the historical sequence of students and exercises on the prediction results.…”

Section: Related Workmentioning

confidence: 99%

“…DNN-based methods, such as NCDM [Wang et al 2020] and CDGK [Wang et al 2021], are then designed for knowledge tracing. In these models, a student and related exercises are mapped into feature vectors based on the student's performance at the current time step.…”

Section: Introductionmentioning

confidence: 99%

Deep Attentive Model for Knowledge Tracing

Wang

Chen

Zhang

2023

AAAI

View full text Add to dashboard Cite

Knowledge Tracing (KT) is a crucial task in the field of online education, since it aims to predict students' performance on exercises based on their learning history. One typical solution for knowledge tracing is to combine the classic models in educational psychology, such as Item Response Theory (IRT) and Cognitive Diagnosis (CD), with Deep Neural Networks (DNN) technologies. In this solution, a student and related exercises are mapped into feature vectors based on the student's performance at the current time step, however, it does not consider the impact of historical behavior sequences, and the relationships between historical sequences and students. In this paper, we develop DAKTN, a novel model which assimilates the historical sequences to tackle this challenge for better knowledge tracing. To be specific, we apply a pooling layer to incorporate the student behavior sequence in the embedding layer. After that, we further design a local activation unit, which can adaptively calculate the representation vectors by taking the relevance of historical sequences into consideration with respect to candidate student and exercises. Through experimental results on three real-world datasets, DAKTN significantly outperforms state-of-the-art baseline models. We also present the reasonableness of DAKTN by ablation testing.

show abstract

Using Knowledge Concept Aggregation towards Accurate Cognitive Diagnosis

Cited by 13 publications

References 26 publications

Enhancing Cognitive Diagnosis Using Un-interacted Exercises: A Collaboration-Aware Mixed Sampling Approach

Enhancing Cognitive Diagnosis Using Un-interacted Exercises: A Collaboration-Aware Mixed Sampling Approach

Cognitive Diagnosis for Programming Domains

Deep Attentive Model for Knowledge Tracing

Contact Info

Product

Resources

About