Knowledge tracing (KT) refers to the problem of predicting students' future performance given their past performance. Scrutinizing previous studies, we can summarize a common learn-to-predict paradigm: a KT model first learns the student's latent knowledge states from historical question-solving learning interactions and then directly predicts whether the student could correctly answer new questions. Alongside the paradigm, existing KT models are dedicated to tailoring refinements for improving predictive performance. However, this has led to increasing model complexity and reduced usability. Inspired by the diagnosis process of human teachers, they conduct correctness prediction based on the students' responses, which are further derived from their latent knowledge states. To achieve this, we propose a novel plug-in Guided diffusiOn mODule (GOOD), which reframes the KT problem as a learn-generate-to-predict paradigm. Specifically, we first employ an existing KT backbone to learn the student's evolving latent knowledge states, subsequently feeding these into our GOOD. Next, GOOD employs a person-wise noise scheduling strategy to add noise to the target responses in the diffusion process, thereby exploring the underlying distribution of response space. Then, GOOD designs a flexible transformer-modulated denoising network to generate target responses utilizing the latent knowledge states as conditional guidance in the reverse process. Finally, the generated responses can explicitly reflect the student's performance, thereby facilitating the correctness prediction. Extensive experiments on four datasets have verified the effectiveness of GOOD in boosting existing KT models to achieve state-of-the-art performance, as well as its generalizability as a flexible plugin.

The effectiveness upper bound of retrieval-augmented generation (RAG) is fundamentally constrained by the semantic integrity and information granularity of text chunks in its knowledge base. Moreover, domain documents are characterized by dense terminology and strong contextual dependencies, which exacerbate the semantic fragmentation of text chunks, thereby making it difficult to efficiently utilize their key information. To address these challenges, this paper proposes QChunker, which restructures the RAG paradigm from retrieval-augmentation to understanding-retrieval-augmentation. Firstly, QChunker models the text chunking as a composite task of text segmentation and knowledge completion to ensure the logical coherence and integrity of text chunks. Drawing inspiration from Hal Gregersen's ''Questions Are the Answer'' theory, we design a multi-agent debate framework comprising four specialized components: a question outline generator, text segmenter, integrity reviewer, and knowledge completer. This framework operates on the principle that questions serve as catalysts for profound insights. Through this pipeline, we successfully construct a high-quality dataset of 45K entries and transfer this capability to small language models. Additionally, to handle long evaluation chains and low efficiency in existing chunking evaluation methods, which overly rely on downstream QA tasks, we introduce a novel direct evaluation metric, ChunkScore. Both theoretical and experimental validations demonstrate that ChunkScore can directly and efficiently discriminate the quality of text chunks. Furthermore, during the text segmentation phase, we utilize document outlines for multi-path sampling to generate multiple candidate chunks and select the optimal solution employing ChunkScore. Extensive experimental results across four heterogeneous domains exhibit that QChunker effectively resolves aforementioned issues by providing RAG with more logically coherent and information-rich text chunks. Notably, this study also establishes a small-domain QA dataset concerning hazardous chemical safety, which fully reveals the significant value of RAG in specialized domains and the generalization capability of the QChunker framework.

As a globally celebrated sport, soccer has seen its appeal greatly amplified by engaging and vivid commentary. Recently, Multi-Modal Large Language Models (MLLMs) have attracted attention in generating soccer commentaries due to their remarkable capacities of understanding different modalities of the input videos. Most of these methods have shown that the use of multiple modalities can enhance the commentary quality, which includes video, audio, and structured meta-data. However, delivering precise and rich commentary requires the ability to accurately discern sub-tle differences in similar backgrounds, events, and players. This presents a significant challenge for existing MLLMs. So we propose SoccerComment, a framework for generating soccer commentary that integrates MLLMs with Retrieval-Augmented Generation (RAG) strategies. This framework enhances inference efficiency and reduces the need for continuous training through a multimodal clustering memory unit and retrieval-augmented in-context learning mechanisms, ultimately improving the accuracy and diversity of the commentary. Based on similar retrieved scenarios, SoccerComment demonstrates outstanding zero-shot performance, offering a new direction and scalable solution for future research in soccer commentary generation.

Knowledge tracing is a core task in intelligent education, aiming to model students’ knowledge states based on their learning behaviors and dynamically predict their mastery of specific concepts. In practical applications, online learning platforms typically provide a large number of questions, but each student can only interact with a small subset. This limited interaction results in extreme data sparsity for certain questions, hindering the model’s ability to build effective representations and make accurate predictions for them. Although recent hypergraph neural network-based knowledge tracing methods can model high-order heterogeneous relationships between questions, improving question representation to some extent, the hypergraph structure often overlooks latent global structural information. This limitation weakens the comprehensive semantic representation of questions, thereby affecting prediction performance. To address these challenges, we propose a Hypergraph Structure-Enhanced for Knowledge Tracing with Gumbel-Softmax sampling (HSEKT-GS). First, we construct a question–concept hypergraph and its dual graph, and incorporate a structural embedding mechanism to capture local high-order relational information between questions and between concepts. Second, to further enhance question representation, we introduce a hypergraph star expansion and use Gumbel-Softmax sampling to generate multiple perturbed embeddings per node to explore structural uncertainty. Finally, the updated representations of all sampled paths are averaged to reveal latent structural links and mitigate the over-smoothing issue in fully connected graphs. In addition, we incorporate a hypergraph structure regularization term as structural supervision to improve the robustness and interpretability of the framework. Experimental results on four publicly available datasets demonstrate that HSEKT-GS outperforms existing baseline methods.

Knowledge tracing (KT) aims to predict students’ responses to given questions based on their historical question-answering interactions. Recent studies have proposed multiple types of KT models, mainly relying on learners’ feedback to capture the evolution of their knowledge states. However, these models leave the influence of learners’ guess behaviors out of consideration. These behaviors would cause unreliable feedback and mislead the inference process of the KT models. Exploring the guess behaviors is important since it can potentially help us understand realistic learning interactions for more accurate knowledge state estimations. In this paper, we propose GuessKT to better trace the learning progress by introducing the guess behaviors into learning feedback attribution, leading to improved prediction performance. Specifically, a windowed attention network is proposed to capture rich information from local interactions, which enhances the reliability of extracted historical feedback. Furthermore, a recovery network is proposed to recover the students’ responses after additional mask processing, which bolsters the model’s ability to recognize guess behaviors. Experiments on five datasets show that our model GuessKT advanced predictive performance over other baselines.

In intelligent systems, knowledge tracing (KT) plays a vital role in providing personalized education. Existing KT methods often rely on students' learning interactions to trace their knowledge states by predicting future performance on the given questions. While deep learning-based KT models have achieved improved predictive performance compared with traditional KT models, they often lack interpretability into the captured knowledge states. Furthermore, previous works generally neglect the multiple semantic information contained in knowledge states and sparse learning interactions. In this paper, we propose a novel model named CAKT that couples contrastive learning with attention networks for interpretable knowledge tracing. Specifically, we use three attention-based encoders to model three dynamic factors of the Item Response Theory (IRT) model, based on designed learning sequences. Then, we identify two key properties related to the knowledge states and learning interactions: consistency and separability. We utilize contrastive learning to incorporate the semantic information of the above properties into the representations of knowledge states and learning interactions. With the training goal of contrastive learning, we can obtain more representative representations of them. Extensive experiments demonstrate the excellent predictive performance of CAKT and the positive effects of considering the two properties. Additionally, CAKT can exhibit high interpretability for captured knowledge states.