Debo Cheng

In recommender systems, popularity and conformity biases undermine recommender effectiveness by disproportionately favouring popular items, leading to their over-representation in recommendation lists and causing an unbalanced distribution of user-item historical data. In this research, a causal graph is constructed to address biases and describe the abstract data generation mechanism. The causal graph is then used as a guide to develop a novel Debiased Contrastive Learning framework for Mitigating Dual Biases, referred to as DCLMDB. In the DCLMDB, both popularity bias and conformity bias are handled during the model training process through contrastive learning, ensuring that user choices and recommended items are not unduly influenced by conformity or popularity. Extensive experiments on three real-world datasets, Movielens-10M, Netflix and Amazon-Art, demonstrate that DCLMDB effectively reduces dual biases. Specifically, on the Movielens-10M dataset, DCLMDB achieves a 35.11% improvement in Recall@20 compared to the backbone model, and significantly outperforms the second-best baseline.

Hallucinations in large language models (LLMs) present major obstacles to reliability in knowledge‐intensive and reasoning‐based tasks. While recent research has explored detection and correction techniques, a unified interpretive framework remains lacking. This survey addresses hallucination mitigation through the lens of explainability, proposing a taxonomy that distinguishes between internal explainability and post hoc explainability. We examine techniques such as attribution tracing, reasoning path construction, and prompt‐based verification, highlighting their roles in transparent diagnosis and structured control. Furthermore, we discuss the constructive role of hallucinations in creative and user‐driven applications, suggesting that context‐aware management may be more effective than blanket suppression. By synthesizing current advances, this review advocates for explainability as a foundation for trustworthy, controllable, and interpretable LLM systems. This article is categorized under: Fundamental Concepts of Data and Knowledge > Explainable AI Technologies > Artificial Intelligence Application Areas > Science and Technology

Cognitive diagnostics in the Web-based Intelligent Education System (WIES) aims to assess students' mastery of knowledge concepts from heterogeneous, noisy interactions. Recent work has tried to utilize Large Language Models (LLMs) for cognitive diagnosis, yet LLMs struggle with structured data and are prone to noise-induced misjudgments. Specially, WIES's open environment continuously attracts new students and produces vast amounts of response logs, exacerbating the data imbalance and noise issues inherent in traditional educational systems. To address these challenges, we propose DLLM, a Diffusion-based LLM framework for noise-robust cognitive diagnosis. DLLM first constructs independent subgraphs based on response correctness, then applies relation augmentation alignment module to mitigate data imbalance. The two subgraph representations are then fused and aligned with LLM-derived, semantically augmented representations. Importantly, before each alignment step, DLLM employs a two-stage denoising diffusion module to eliminate intrinsic noise while assisting structural representation alignment. Specifically, unconditional denoising diffusion first removes erroneous information, followed by conditional denoising diffusion based on graph signal to eliminate misleading information. Finally, the noise-robust representation that integrates semantic knowledge and structural information is fed into existing cognitive diagnosis models for prediction. Experimental results on three publicly available web-based educational platform datasets demonstrate that our DLLM achieves optimal predictive performance across varying noise levels, which demonstrates that DLLM achieves noise robustness while effectively leveraging semantic knowledge from LLM.