Hui Zhou

Consumer reviews provide valuable post-hoc evidence for detecting false advertising in live-streaming e-commerce. However, detecting unseen or emerging deceptive promotions remains challenging due to noisy reviews, incomplete context, and cross-category variation. To address these challenges, we propose DARE-Net, a Domain-Adaptive Rationale Evolution Network for cross-domain false advertising detection from consumer reviews. Instead of directly transferring general semantic or topical patterns, DARE-Net formulates the task as rationale-centric discrepancy evidence learning. Specifically, a hierarchical discrepancy-aware rationale mining module extracts multi-granularity deception-related rationales from noisy reviews; a domain-adaptive rationale transformation module disentangles domain-specific bias from transferable deception evidence; and an evidence-boundary decision module adaptively aggregates decisive rationales with class-adaptive margins to alleviate imbalance-induced boundary bias. Experiments on multi-domain live-streaming e-commerce review datasets show that DARE-Net consistently outperforms state-of-the-art cross-domain and task-related detection methods, achieving a relative improvement of 2.52% over the best-performing baseline. Further analysis demonstrates that DARE-Net can capture transferable deception evidence and provide interpretable predictions across diverse product domains, indicating its effectiveness for robust false advertising detection in dynamic live-streaming e-commerce environments.

Recent advances in prompt learning have opened new avenues for enhancing natural language understanding in domain-specific tasks, including code vulnerability detection. Motivated by the limitations of conventional binary classification methods in capturing complex code semantics, we propose a novel framework that integrates a two-stage prompt optimization mechanism with hierarchical representation learning. Our approach leverages graphon theory to generate task-adaptive, structurally enriched prompts by encoding both contextual and graphical information into trainable vector representations. To further enhance representational capacity, we incorporate the pretrained model CodeBERTScore, a syntax-aware encoder, and Graph Neural Networks, enabling comprehensive modeling of both local syntactic features and global structural dependencies. Experimental results on three public datasets—FFmpeg+Qemu, SVulD and Reveal—demonstrate that our method performs competitively across all benchmarks, achieving accuracy rates of 64.40%, 83.44% and 90.69%, respectively. These results underscore the effectiveness of combining prompt-based learning with graph-based structural modeling, offering a more accurate and robust solution for automated vulnerability detection.

API recommendation has been widely used to enhance developers’ efficiency in software development. However, existing API recommendation methods for dynamic languages such as Python usually suffer from the limitations of incorrect type inference and lack of rich contextual semantics. To address these issues, we propose in this paper a novel approach, PyBartRec, to recommend APIs for Python programs concerning their rich semantics. Instead of analyzing the data flow information only, our approach utilizes a Transformer-based pre-trained model to extract the semantic features of Python code snippets. Such contextual information allows our approach to recommend correct APIs even when the APIs are not included in the local data flow information. We also use such information to perform a post-processing type inference. By narrowing the range of candidate types, our approach can recommend APIs accurately even in the failure scenarios of type inference. We evaluated PyBartRec in eight popular Python projects and the experimental results show that our approach significantly outperforms the state-of-the-art solutions. In particular, the average top-1 accuracy and average top-10 accuracy of PyBartRec within the same project are over 40%, and 60%, respectively. In cross-project recommendation, the MRR of PyBartRec is also over 40%.

With the rapid development of ocean transportation, vessel trajectory mining and analysis have become more and more important. Trajectory clustering is the basis of practical applications. Traditional trajectory similarity calculation methods are relatively low in accuracy and efficiency, the existing deep learning methods mostly extract features of static ones, ignoring the multi-feature combination of dynamic and static features. This paper proposes a static-dynamic-feature fusion method based on convolutional auto-encoder, which can extract more perfect trajectory features in artificial intelligence field. Experiments on real datasets have demonstrated that compared with other methods, there is at least a 5%-10% improvement in terms of classic clustering metrics and lower time consumption.