华东政法大学

Legal reasoning is a fundamental component of legal analysis and decision-making. Existing computational approaches to legal reasoning predominantly rely on generic reasoning frameworks such as syllogism, which do not comprehensively examine the nuanced process of legal reasoning. Moreover, current research has largely focused on criminal cases, with insufficient modeling for civil cases. In this work, we present a novel framework to explicitly model legal reasoning in the analysis of Chinese tort-related civil cases. We first operationalize the legal reasoning process in tort analysis into the three-module LexChain framework, with each module consisting of multiple finer-grained sub-steps. Informed by the LexChain framework, we introduce the task of tort legal reasoning and construct an evaluation benchmark to systematically assess the critical steps within analytical reasoning chains for tort analysis. Leveraging this benchmark, we evaluate existing large language models for their legal reasoning ability in civil tort contexts. Our results indicate that current models still fall short in accurately handling crucial elements of tort legal reasoning. Furthermore, we introduce several baseline approaches that explicitly incorporate LexChain-style reasoning through prompting or post-training. The proposed baselines achieve significant improvements in tort-related legal reasoning and generalize well to related legal analysis tasks, demonstrating the value of explicitly modeling legal reasoning chains to enhance the reasoning capabilities of language models.

While the Biodiversity Beyond National Jurisdiction (BBNJ) Agreement establishes a comprehensive environmental impact assessment (EIA) regime, its application of the EIA provisions to the international seabed area (the Area) presents significant challenges. First, International Seabed Authority (ISA) Member States have differing ratification and signature statuses regarding the BBNJ Agreement, leading to fragmented and asymmetrical obligations. Second, normative ambiguity surrounds the “not undermine” proviso, and an institutional disconnection exists between the Clearing-House Mechanism under the BBNJ Agreement and the ISA. Third, the “equivalence exception” under Article 29(4) of the BBNJ Agreement generates a critical interpretive dilemma. The absence of institutional mechanisms for determining equivalence risks unilateral self-assessment and forum-shopping. Furthermore, substantive disparities exist between the BBNJ Agreement and ISA regulations, particularly regarding the screening process, Cumulative Environmental Effects, and Strategic Environmental Assessment, alongside asymmetries in procedural transparency and stakeholder engagement. Finally, an empirical analysis reveals that the EIA practices in the Area remain scarce and inaccessible, and existing Environmental Impact Statements appear not fully aligned with the standards prescribed by Article 33(2) of the BBNJ Agreement. Therefore, this article puts forward recommendations to optimize the application of the EIA provisions under the BBNJ Agreement. First, concerted efforts must be made to advance the ratification and accession of the BBNJ Agreement by ISA Member States. Second, the “not undermine” proviso should be interpreted narrowly to ensure the BBNJ Agreement’s implementation does not actively dismantle the effectiveness of the ISA. Third, Parties should facilitate the determination of equivalence through harmonization and scientific advice, leveraging the supplementary role of the Scientific and Technical Body. Fourth, cooperation must be formalized through institutional frameworks, specifically by signing a Memorandum of Understanding and enhancing inter-institutional expert consultation. Finally, stakeholders must enhance EIA practices by standardizing the mechanism for updating reports and ensuring that dual-member States comply with the standards of both regimes.

BACKGROUND:Child maltreatment is highly prevalent in rural areas, yet evidence-based parenting programs remain largely inaccessible because of caregivers' inflexible work schedules and the shortage of trained facilitators. To enhance its scalability and cost-effectiveness, we adapted the Chinese version of Parenting for Lifelong Health for Young Children (PLH-YC) program into a self-directed online format with two group discussions and evaluated its impact in rural China. METHODS:Twelve classes (191 caregivers) in one rural kindergarten were cluster-randomized 1:1 to either a self-directed online PLH-YC program (100 caregivers) or a book-reading control (91 caregivers). Assessments were conducted at baseline, two weeks post-intervention, and three-month follow-up. Analyses adhered to the intention-to-treat principle, using zero-inflated negative binomial regression for count outcomes (e.g., child maltreatment) and linear mixed-effects regression for continuous outcomes (e.g., positive parenting practices). RESULTS:Two weeks post-intervention, caregivers in the self-directed group reported a 33% reduction in overall child maltreatment (IRR = 0.67, p < 0.001), a moderate increase in positive parenting (d = 0.53, p < 0.001), a small-to-moderate changes in parent-child relationship quality (d = 0.32, p < 0.001) and parenting stress (d = -0.31, p < 0.001) relative to controls. At the three-month follow-up, reductions in child behavior problems became significant (d = -0.40, p < 0.001). Parenting efficacy and parental depressive symptoms did not differ between groups at both time points. CONCLUSION:The self-directed online PLH-YC program, combined with group discussions, provided an effective, low-cost, and scalable option for disadvantaged rural caregivers in China.

In 2023, the Commission of Small Island States (COSIS) submitted a request to the International Tribunal for the Law of the Sea (ITLOS) seeking an advisory opinion on state obligations under the UNCLOS in relation to climate change. The request aimed to address the existential threats posed to the Small Island Developing States (SIDS), including rising sea levels, extreme weather, and the degradation of marine ecosystems. On May 21, 2024, the ITLOS delivered a groundbreaking advisory opinion on the protection of the marine environment from pollution linked to climate change (case no. 31). This article will first outline the background and key aspects of the advisory opinion, followed by an analysis of relevant case law. It will critically assess state obligations under the UNCLOS in the context of the opinion and explore its broader implications for global climate change mitigation. Additionally, the article will consider how the opinion emphasizes the necessity of international cooperation in tackling climate challenges, while also addressing the political and practical barriers that could limit its implementation. Despite these challenges, the article will offer recommendations on the advisory opinion’s significance and its potential impact on international climate change efforts.

Unsupervised anomalous sound detection aims to learn acoustic features solely from the operational sounds of normal equipment and identify potential anomalies based on these features. Recent self-supervised classification frameworks based on machine ID metadata have achieved promising results, but they still face two challenges in industrial acoustic scenarios: Log-Mel spectrograms tend to weaken high-frequency details, leading to insufficient spectral characterization, and when normal sounds from different machine IDs are highly similar, classification constraints alone struggle to form clear intra-class structures and inter-class boundaries, resulting in false positives. To address these issues, this paper proposes FGASpecNet, an anomaly detection model integrating spectral enhancement and frequency-gated attention. For feature modeling, a spectral enhancement branch is designed to explicitly supplement spectral details, while a frequency-gated attention mechanism highlights key frequency bands and temporal intervals conditioned on temporal context. Regarding loss design, a joint training strategy combining classification loss and metric learning loss is adopted. Multi-center prototypes enhance intra-class compactness and inter-class separability, improving detection performance in scenarios with similar machine IDs. Experimental results on the DCASE 2020 Challenge Task 2 for anomalous sound detection demonstrate that FGASpecNet achieves 95.04% average AUC and 89.68% pAUC, validating the effectiveness of the proposed approach.