Risk Analysis最新文献

Failure event narratives contain detailed and valuable information describing how failures initiate and propagate. Event causality analysis can help improve the understanding of failure physics and facilitate the use of non-failure data (e.g., near-misses and degradations) to complement the limited data pool of failures, which is common in high-reliability industries such as the nuclear power industry. Automatically extracting event causality from text data, however, is challenging given complex and diverse language structures and causal patterns, and the lack of access to large, annotated datasets for use as training data. Existing automated mining approaches are mainly knowledge-based and extract causality using a set of predefined keywords and rules, which have difficulty achieving good performance. In this paper, we propose a novel large language model (LLM)-based approach for automated causality extraction. It leveraged the strong capability of LLM to understand intricate language patterns in long-range contexts and accurately extract cause-and-effect pairs from texts. The proposed approach has a twofold framework: causality detection and causality extraction. The causality detection step trained a deep learning model to identify texts with causality. The causality extraction step developed a T5-CE LLM to identify and extract cause-and-effect pairs in each text sample. A large, annotated dataset of the U.S. nuclear power plant failure event reports was used to train and evaluate the models. The model evaluation was performed using three performance metrics, including precision, recall, and F₁ score. The proposed approach can effectively detect implicit and embedded causalities across multiple sentences.

In seismic probabilistic safety assessment (SPSA), ongoing efforts to reduce core damage frequency (CDF) encounter fundamental limitations due to modeling saturation, epistemic uncertainties, and the diminishing contribution of extremely low-probability events. This paper introduces and develops the concept of "inherent CDF"-a residual risk floor that persists when further reductions in CDF become practically or physically unattainable despite best-estimate modeling and conservative design. Drawing from recent literature, regulatory philosophy, and structural fragility modeling, the paper synthesizes the rationale for formalizing this floor within seismic PSA methodology. A conceptual framework is proposed to identify, justify, and declare the inherent CDF based on hazard-fragility convolution, uncertainty bounds, and plateauing risk behaviour at high ground motions. The implications for licensing, risk-informed decision-making, and safety optimization are examined, along with potential extensions to multi-hazard PSA domains. This study advocates for the integration of residual risk acknowledgment into modern safety frameworks, providing a technically grounded and transparent foundation for defining acceptable seismic risk in nuclear power plants.

We have analyzed 39 major ammonium nitrate (AN) accidents (1916-2022) that reveal systemic catastrophic risks have not diminished over the past century. This challenges the prevailing narrative of improved process safety of AN on account of iterative guidelines and technology advances. Applying extreme value theory (EVT), we model temporal occurrence as a homogeneous Poisson process (maximum likelihood estimate λ ≈ 0.23 events/year) and casualty severity as heavy-tailed (generalized Pareto distribution, shape parameter ξ ≈ 1.13). These findings indicate that there are persistent systemic patterns. This is shown by the implied "shadow mean" that exceeds the sample mean by 1.5-19-fold, and societal risk profiles occupy the "intolerable" region per frequency-number curve criteria. To address this problem, we establish an EVT-based systemic monitoring framework with quantitative benchmarks, identifying 50-year (∼315 fatalities) and 100-year (∼715 fatalities) return levels as a data-driven baseline for assessing safety performance. This ensemble framework provides a necessary tool for stakeholders to monitor realized systemic risks over mechanism-driven simulations or narrative expectations.

People's risk perceptions are crucial for climate change adaptation, influencing individual decisions and policy effectiveness. Although many studies highlight the importance of social influences and social norms in this context, the mechanisms through which they shape individual risk perceptions and adaptation behavior remain unclear. To address this gap, we analyze cross-country survey data (N = 1612) from coastal areas in the Netherlands, United Kingdom, and the USA with a focus on flood risk and adaptation behavior. Our statistical analysis reveals several important patterns in social interactions, and the ways in which these social interactions influence individual risk perceptions. First, we find limited social engagement regarding risks and adaptation, with a significant portion of respondents (50%) reporting no interactions with peers on these topics. Among those who do engage, social interactions on flood risk and adaptation appear infrequent (fewer than five times per year). Second, contrary to common assumptions, individuals who discuss flood risk and adaptation, rarely do so with neighbors. Moreover, homophily-shared socio-demographic characteristics-is not the primary determinant of who interacts on the topic. Third, we see that those with hazard experience and those with higher risk perceptions are more likely to interact with others on the topics of these risks and climate adaptation, confirming that social amplifications might be in place. These findings provide unique insights into the social dynamics underlying the evolution of individual risk perceptions, offering the potential to refine models of social influence in climate change and social tipping points. They also highlight potential synergies between communication strategies and policy tools to support timely and, possibly transformational, adaptation.

The systemic diffusion of disinformation on social media poses escalating threats to digital societies, causing cognitive distortion, financial instability, and public crises that demand precision governance. Existing models exhibit critical limitations. For example, epidemiological approaches neglect individual decision heterogeneity, game-theoretic frameworks assume complete rationality, and cognitive psychology paradigms lack cross-scale risk coupling mechanisms. We propose a novel disinformation propagation model which incorporates (a) individual-level dynamics via risk perception thresholds, decision sensitivity, and cognitive inertia; (b) network-level equations that formalize topology-driven risk cascades; and (c) intervention analytics that quantify minimum effective intensity thresholds. Specifically, we do the following main work: (1) We construct a network propagation model without authoritative intervention and prove equilibrium existence while it is not unique; (2) we develop a dual-index system of nodal risk exposure and vulnerability to identify critical superspreaders; (3) we formulate an intervention model combining cognitive correction with propagation suppression, prove equilibrium existence under interventions, and derive sufficient conditions for unique convergence; (4) we establish the bounded intervention efficacy theorem, ensuring predictable outcomes when intervention intensity exceeds critical thresholds; (5) we derive lower bounds on convergence time and compare convergence rates between intervention and nonintervention scenarios; and (6) we quantify how individual heterogeneity and intervention intensity jointly modulate systemic risks. These findings provide a comprehensive theoretical framework for constructing a disinformation immune system. By clarifying the coupling dynamics between propagation mechanisms and intervention strategies, our research offers quantifiable decision-making tools for digital governance.

The Emissions Trading System (ETS), modeled after the Kyoto Protocol, is widely recognized for its role in advancing the sustainable development goals (SDGs). However, this market-based mechanism has been criticized from a Keynesian perspective, which highlights that investment inertia in technology and climate governance can impede effective climate risk management. To address this debate, this study constructs a novel conceptual framework integrating institutional theory and the resource-based view. Employing a staggered difference-in-differences (DID) design with global data from 2000 to 2023, we empirically examine the complex relationship between ETS and physical climate risk. The results indicate that: First, the ETS, as a substantive climate governance tool driven by public pressure, significantly reduces physical climate risks, particularly acute climate risks. Second, the ETS governs climate risk primarily through three pathways-climate mitigation, adaptation, and finance-while also exhibiting spillover effects and threshold characteristics. Additionally, the effectiveness of the ETS is influenced by ideological and economic alignments, with notable variations among capitalist, non-EU, and non-OECD countries. Finally, high carbon prices, taxes, and allowances cause imbalances in the carbon market. This study not only provides a comprehensive explanation of the underlying mechanisms through which the ETS affects physical climate risk but also offers theoretical insights and empirical support for ETS optimization and the design of other climate policies.

Artificial intelligence (AI) is set to transform traditional healthcare delivery and patient care. However, this transformation presents a range of challenges in medical AI decision-making. To address these challenges, this study aims to develop a risk analysis framework for medical AI. A mixed-methods approach is adopted, combining quantitative and qualitative analyses. First, Latent Dirichlet Allocation (LDA) topic modeling is used to analyze 1618 news articles from a leading health information website in China, and this identifies eight risk attributes (i.e., privacy and security, social bias, decision-making status, data quality, algorithm inference, system performance, liability attribution, and regulatory legislation). Second, to identify the risk attributes of medical AI from the patient's perspective, semi-structured interviews with 21 patients and analysis using grounded theory were employed, identifying six key risk attributes of concern to patients from the initial set of eight. Lastly, an online survey of 396 patients was conducted, followed by a Choice-Based Conjoint (CBC) analysis to assess patient preferences in relation to these six risk attributes. The results show that patients prioritize risks in the following order: Data quality (30.20%), privacy and security (29.50%), social bias (19.10%), system performance (13.70%), liability attribution (6.87%), and algorithm inference (0.59%). This study proposes a risk analysis framework that offers practical insights for healthcare policymakers, medical AI developers, and risk analysts.

Designing effective food safety monitoring schemes is a complex task involving multiple, often conflicting, criteria. This study applied Multi-Criteria Decision Analysis (MCDA) to evaluate and identify optimal aflatoxin monitoring schemes along a Dutch dairy supply chain-a critical context where aflatoxin B1 (AFB1) contamination in feed can lead to aflatoxin M1 (AFM1) in milk, potentially posing public health concerns and economic losses. Monitoring schemes differed in detection intensity at feed mills and dairy farms, defined as the probability of identifying contaminated batches (low: 50%, medium: 80%, high: 90%) and determined by the number of monitoring batches and corresponding sample sizes used for AFB1/AFM1 sampling and analysis. Performance scores for each monitoring scheme were derived from quantitative models, scientific evidence, and expert consultation, while preference weights for criteria were elicited separately from representatives of the feed industry, dairy industry, and from a combined supply chain perspective. Results revealed that all stakeholder groups prioritized public health, but differed in their weighting of monitoring costs, production losses, customer trust, and implementation complexity. The feed industry preferred high-intensity detection at both control points, while the dairy industry preferred medium-intensity at feed mills and high-intensity at farms. Overall, the MCDA framework facilitated a transparent and evidence-based approach to identify an optimal monitoring scheme, highlighting the importance of stakeholder engagement in designing programs that are not only scientifically robust but also socially responsive and aligned with the WHO Global Strategy for Food Safety and the Sustainable Development Goals.