Dynamic Key-Value Memory Networks for Knowledge Tracing

By Jiani Zhang, Xingjian Shi, Irwin King, Dit-Yan Yeung. The Web Conference, 2016

863 citations recorded by the source corpus.

Knowledge Tracing (KT) is a task of tracing evolving knowledge state of students with respect to one or more concepts as they engage in a sequence of learning activities. One important purpose of KT is to personalize the practice sequence to help students learn knowledge concepts efficiently. However, existing methods such as Bayesian Knowledge Tracing and Deep Knowledge Tracing either model knowledge state for each predefined concept separately or fail to pinpoint exactly which concepts a student is good at or unfamiliar with. To solve these problems, this work introduces a new model called Dynamic Key-Value Memory Networks (DKVMN) that can exploit the relationships between underlying concepts and directly output a student's mastery level of each concept. Unlike standard memory-augmented neural networks that facilitate a single memory matrix or two static memory matrices, our model has one static matrix called key, which stores the knowledge concepts and the other dynamic matrix called value, which stores and updates the mastery levels of corresponding concepts. Experiments show that our model consistently outperforms the state-of-the-art model in a range of KT datasets. Moreover, the DKVMN model can automatically discover underlying concepts of exercises typically performed by human annotations and depict the changing knowledge state of a student.

Subfields: Artificial Intelligence, Computer Science Applications

View Paper: https://www.semanticscholar.org/paper/c96bdf1fe9a43e0a1af9459b9572eb1f69527755

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Science Applications

Citations in the corpus

Amalgamating Knowledge from Two Teachers for Task-oriented Dialogue System with Adversarial Training (2020, background): “…Memory and KB memory Different from Mem2Seq (Madotto et al., 2018), our dialogue memory is implemented with a dynamic key-value memory network (Zhang et al., 2017), which maintains a timely updated key memory to keep track of attention history and a ﬁxed value-memory to store the dialogue…”
Variational Item Response Theory: Fast, Accurate, and Expressive (2020, methodology): “We also compare our deep generative IRT models with the purely deep learning approach called Deep-IRT [47] (see Sec.”
A Self Attentive model for Knowledge Tracing (2019, methodology): “We adopted the processed data from [11].”
Sequential Recommendation with User Memory Networks (2018, background): “The memory vector is left unchanged if either the weight or the erase signal is zero [32].”
MathBERT: A Pre-trained Language Model for General NLP Tasks in Mathematics Education (2021, methodology): “‘AUC’ and ‘ACCU’ are used to measure Tkt because historically they are evaluated by both metrics [14, 20, 25, 33].”
Finding Similar Exercises in Online Education Systems (2018, background): “For instance, in E1 of 1Concepts (short for knowledge concepts [35, 46] or knowledge points [4]) are previously labeled (e.g. in ASSISTments [4, 46]) or can be easily obtained by algorithms [26].”
Context-Aware Attentive Knowledge Tracing (2020, methodology): “R, DKVMN, DKVMN-R, SAKT, and SAKT-R on the ASSISTments2009 and 2017 datasets where question IDs are available; DKT-R, DKT+-”
Fine-Grained Similarity Measurement between Educational Videos and Exercises (2020, background): “Generally, similar videos and exercises are those having common concepts (short for knowledge concepts [26, 45] or knowledge points [15]).”
Knowledge Tracing: A Survey (2022, methodology): “Publication date: February 2023. such as DKT which uses a hidden state to model knowledge states over a single KC, to later works by memory-augmented KT models (e.g., DKVMN and SKVMN) and by text-aware KT models (e.g., EKT) which use matrices to model knowledge states over multiple KCs, a trend in d”
Addressing two problems in deep knowledge tracing via prediction-consistent regularization (2018, methodology): “We have adopted the processed data provided by [21] with a train / test split of ratio 70:30, and exercise tags are used.”
GaAN: Gated Attention Networks for Learning on Large and Spatiotemporal Graphs (2018, background): “Neural attention mechanism Neural attention mechanism is widely adopted in deep learning literature and many variants have been proposed (Xu et al., 2015b; Seo et al., 2017; Zhang et al., 2017; Vaswani et al., 2017; Cheng et al., 2017; Zhang et al., 2018).”