Evidence prediction is the cornerstone of litigation, which requires complex legal reasoning to bridge the gap between facts and the parties’ assertions. However, most existing large language model (LLM) benchmarks do not offer a systematic assessment for evidence planning. Therefore, we introduce CLER, a large-scale Chinese Litigation Evidence Reasoning benchmark. CLER is built through a four-stage pipeline from millions of real civil and criminal judgments. The pipeline includes (1) semi-structured data collection, (2) LLM-based evidence extraction, (3) purpose refinement through decomposition into legal elements, and (4) validation through a tri-party adversarial review method that imitates courtroom debate. To this end, we obtain over 100,000 samples and conduct extensive experiments on 23 baselines, including open-weight, commercial, domain-specific LLMs and retrieval-augmented generation (RAG) systems. The findings indicate that the task poses a major challenge for current LLMs. RAG strategies, particularly a hybrid method retrieving similar cases, achieve the best performance with an F1 score of 54.4%, surpassing state-of-the-art proprietary models and improving the base model by at least 22%. Our analysis reveals several key insights. There is a significant performance gap with an average difference F1 of 16.2% exists between element checking in civil cases and narrative completeness in criminal cases. There is an inverted U-shaped correlation between case length and model performance, as insufficient or overloaded information leads to abandonment of precise reasoning, causing an average drop of approximately 10% in F1.

Online social networks (OSNs) have become the most popular platforms for information creation and sharing, actively encouraging users to showcase personal data such as profiles, opinions, and activities. The paradox of sharing and privacy issues is a big trouble for users. Most privacy protection services provided for users, such as access control, need to be manually configured to follow the privacy-aware user intents. It causes a significant cognitive burden for users, due to the dynamic changes in user relationships and the existence of many malicious users in the network. An adaptive privacy-aware sharing strategy management framework (APUS) based on multi-agent reinforcement learning (MARL) is proposed, which provides a service for users to adjust their sharing strategies dynamically in the context of potential malicious users. In APUS, each agent observes some users and has similar workloads, and MARL is adopted for learning cooperation. Besides, the optimal sharing strategies are dynamically derived to manage users' privacy settings for shared content while preventing the impact of malicious users. Experiments conducted on real-world datasets show that this work can effectively promote a win-win situation between the OSN platform and users.

Retrieving superior legal articles involves identifying relevant legal articles that hold higher legal effectiveness. This process is crucial in legislative work because superior legal articles form the legal basis for drafting new laws. However, most existing legal information retrieval research focuses on retrieving legal documents, with limited research on retrieving superior legal articles. This gap restricts the digitization of legislative work. To advance research in this area, we propose SLARD: A Chinese Superior Legal Article Retrieval Dataset, which filters 2,627 queries and 9,184 candidates from over 4.3 million effective Chinese regulations, covering 32 categories, such as environment, agriculture, and water resources. Each query is manually annotated, and the candidates include superior articles at both the provincial and national levels. We conducted detailed experiments and analyses on the dataset and found that existing retrieval methods struggle to achieve ideal results. The best method achieved a R@1 of only 0.4719. Additionally, we found that existing large language models (LLMs) lack prior knowledge of the content of superior legal articles. This indicates the necessity for further exploration and research in this field.

Personalized Federated Learning (PFL) meets each user’s personalized needs while still facing the high communication costs due to the large amount of data transmission and frequent communication. Decentralized PFL (DPFL) as an alternative discards the central server in PFL, which reduces the pressure of communication and the risk of server failure by using peer-to-peer communication.Nevertheless, DPFL still suffers from the significant communication pressure due to the transmission of a large number of model parameters, especially numerous nodes. To address the issues, we propose a novel personalized framework, DFedHP, in which each client utilizes a hypernetwork to generate the shared part of model parameters and train the personalized parameters separately. The number of parameters in a hypernetwork is much smaller than those in a typical local network, so hypernetwork aggregation reduces communication costs and the risk of privacy leakage. Furthermore, DFedHP can seamlessly integrate into existing DPFL algorithms as a plugin to boost their efficacy. At last, extensive experiments on various data heterogeneous environments demonstrate that DFedHP can reduce communication costs, accelerate convergence rate, and improve generalization performance compared with state-of-the-art (SOTA) baselines.

The impressive performance of transformer models has sparked the deployment of intelligent applications on resource-constrained edge devices. However, ensuring high-quality service for real-time edge systems is a significant challenge due to the considerable computational demands and resource requirements of these models. Existing strategies typically either offload transformer computations to other devices or directly deploy compressed models on individual edge devices. These strategies, however, result in either considerable communication overhead or suboptimal trade-offs between accuracy and efficiency. To tackle these challenges, we propose a collaborative inference system for general transformer models, termed CoFormer. The central idea behind CoFormer is to exploit the divisibility and integrability of transformer. An off-the-shelf large transformer can be decomposed into multiple smaller models for distributed inference, and their intermediate results are aggregated to generate the final output. We formulate an optimization problem to minimize both inference latency and accuracy degradation under heterogeneous hardware constraints. DeBo algorithm is proposed to first solve the optimization problem to derive the decomposition policy, and then progressively calibrate decomposed models to restore performance. We demonstrate the capability to support a wide range of transformer models on heterogeneous edge devices, achieving up to 3.1× inference speedup with large transformer models. Notably, CoFormer enables the efficient inference of GPT2-XL with 1.6 billion parameters on edge devices, reducing memory requirements by 76.3%. CoFormer can also reduce energy consumption by approximately 40% while maintaining satisfactory inference performance.

With the widespread application of temporal knowledge graph reasoning (TKGR) models, there is an increasing demand to reduce the memory consumption and enhance the reasoning efficiency. Knowledge distillation (KD) is a classical approach to achieve model compression and acceleration, which has been gradually introduced into the TKGR domain. Through KD, the expertise of a high-capability teacher TKGR model can be transferred to a lightweight student TKGR model. The effective transfer of reasoning knowledge primarily faces two major challenges. The first is how to extract high-quality and high-value knowledge from the teacher to the student to achieve better teaching outcomes. The second is how to encourage the teacher to improve the teaching pattern so that the knowledge is more easily assimilated by the student. Motivated by these challenges, this article firstly designs a soft-label evaluation mechanism to mitigate the problem of anomaly diffusion and knowledge transfer redundancy by measuring the confidence and entropy changes of soft labels, then proposes a mutual learning-empowered KD (MLEMKD) framework for compressing TKGR models. It refines the distribution of knowledge by analyzing the cognitive differences between teacher and student models on training samples, which enhances the acceptability of knowledge. Extensive experiments conducted on four benchmark datasets demonstrate that MLEMKD significantly outperforms existing KD methods.