To address the limitations of the Practical Byzantine Fault Tolerant (PBFT) consensus algorithm, this paper proposes several improvement measures. Firstly, design a mechanism for evaluating nodes to calculate the reputation score of nodes, and then classify them according to their reputation scores. Nodes with lower reputation scores can receive consensus outcomes and are not involved in the consensus process. Secondly, a dynamic node joining and exiting mechanism was designed to improve the consensus efficiency of network nodes that cannot join or exit dynamically. Finally, the original PBFT algorithm's three-stage consensus process is improved to decrease the communication times. Experimental results indicate that the refined consensus algorithm has better performance than the original PBFT algorithm in security, the number of communication times, and consensus delay.

Multimodal sentiment analysis jointly leverages textual, visual, and acoustic signals to understand human affect and underpins applications in human–computer interaction, education, and media analytics. In practice, ”pairwise-first, then stacking” fusion is simple and stable but suffers from order sensitivity, a longer inference path, and memory usage that grows rapidly with sequence length. To address these limitations, we propose OPAL-T, a one-shot trimodal interaction framework that performs joint reasoning over language, video, and audio in a single pass. OPAL-T comprises four components: (1) a Low-Rank Tri-modal Interaction (LRTI) kernel that replaces the explicit tensor cube with a small set of rank-R factors, reducing the dominant complexity from cubic to near-quadratic; (2) Gated Dual-Path Symmetric Fusion (GDSF), which stabilizes training by fusing order-swapped key–value paths; (3) Symmetric Single-Path Inference (Sym-LRTI), which collapses GDSF at inference into a parameter-free symmetric path; and (4) Multi-label Prototype Flexible Contrast (MPFC), which structures the multi-label emotion space via Jaccard-weighted relations and EMA-updated class prototypes. Across three public datasets, OPAL-T achieves competitive overall performance compared with state-of-the-art methods.

In federated cross-domain recommendation, the existing research has widely focused on scenarios where at least some users or items overlap. How to enable cross-domain recommendation over non-overlapping users and items has emerged as a challenging research direction. However, current privacy-preserving cross-domain recommendation methods struggle to achieve effective semantic alignment across domains in such scenarios, hindering their ability to capture latent correlations between interaction graphs. Moreover, when refining noisy node representations and interaction sequences, these methods often overlook temporal noise implicitly embedded in semantic information, as well as the potential of jointly leveraging textual and temporal features for collaborative denoising. To address these challenges, this paper proposes a Semantic Federated Clustering-driven Cross-Domain Recommendation model (SFC-CDR). Specifically, we first design a clustering-anchored contrastive learning framework that performs clustering using weighted multi-layer node semantic representations. The weighted aggregated centroids serve as semantic anchors to dynamically update the pool of positive and negative samples in contrastive learning, thereby achieving geometric alignment in the shallow feature space and capturing fine-grained semantic correlations in the implicit concept space. Furthermore, we design a Joint Temporal–Frequency Filtering (JTFF) module. In the frequency domain, features are processed via an MoE adapter and FFT-based filtering, while in the temporal domain, a temporal convolutional network (TCN) is employed to model temporal patterns in user–item interaction sequences. The resulting temporal and frequency-domain representations are integrated through Temporal–Frequency Refinement Gating (TFRG), enabling complementary refinement of node semantic representations and user behavior sequences. Extensive experiments conducted on four real-world datasets demonstrate that SFC-CDR outperforms state-of-the-art baselines.

Graph motifs capture crucial structural patterns within user–item interaction graphs and offer meaningful semantics that are typically underutilised in conventional graph-based collaborative filtering. Existing contrastive learning methods in recommender systems rely primarily on simple perturbations, which limits their ability to leverage deeper motif-based structural information. To address this gap, we propose a novel motif-guided contrastive learning framework for recommender systems. To exploit complementary structural biases inherent to user–item interactions, our approach explicitly incorporates three distinct motifs into the construction of the contrastive view. By contrasting views that capture the structural differences among different motifs, our model learns meaningful group-level relationships and potentially suppresses noise arising from sparse or isolated interactions. Extensive experiments on four real-world recommendation datasets validate that our motif-based contrastive approach achieves the best overall performance, outperforming state-of-the-art baselines on three of the four benchmarks with statistically significant margins on the more skewed datasets, while remaining competitive on the fourth, which demonstrates notable robustness and improved accuracy.