Leveraging contextual events on structure-aware next activity prediction
本文通过引入基于实例图的方法并结合上下文信息，利用图神经网络改进了流程中下一个活动的预测性能。
0 citationsRead paper
Selected work
Representative Papers
Theia: Large-Scale Multimodal Captioning and Automated Validation of the Incidents1M Dataset for Data-Free Distillation
This study addresses the scarcity of high-quality, large-scale multimodal datasets with semantically aligned image-text pairs in disaster response, which hinders data-free knowledge distillation for vision-language models. Building upon the purely visual Incidents1M dataset, the authors leverage the Qwen3.5 series of large language models to generate image captions and introduce an innovative image-blinded LLM-as-a-Judge automated verification mechanism. This approach simulates the modality gap faced by student models during distillation, effectively identifying and suppressing annotation inconsistencies and false positives inherent in human labeling. The method produces high-fidelity descriptions for 100,000 images, achieving a semantic consistency score of 78.65 on 173,179 annotated pairs and a verification accuracy of 77.6%, substantially enhancing the semantic reliability of data-free distillation.
0 citationsRead paper
A forgery attack on the Block.co blockchain-based digital credential certification system
This study addresses a critical vulnerability in existing blockchain-based academic credential verification systems, such as Block.co, which lack a trustworthy mechanism to securely bind an educational institution’s real-world identity to its digital counterpart. The authors present the first practical forgery attack against such systems, exploiting the disconnect between blockchain immutability and identity validation logic to generate fraudulent credentials that are erroneously validated as legitimate. This work exposes a fundamental flaw in decentralized credentialing models—namely, the insecure identity-binding process—and demonstrates that this weakness introduces systemic security risks across similar platforms. The proposed attack methodology is generalizable, highlighting a pervasive threat that undermines the integrity of blockchain-issued digital credentials and offering crucial insights for the secure design of future systems.
0 citationsRead paper
A Two-Stage Decision Support System for Sustainability-Aware Long Short Portfolio Optimization
This study addresses the challenge of constructing long–short portfolios that simultaneously achieve sustainability and superior risk-adjusted returns by integrating environmental, social, and governance (ESG) factors in a market-regime-adaptive manner. The authors propose a two-stage decision framework: first, assets are selected for long and short positions using a hybrid approach combining TODIMSort multi-criteria classification with MEREC-based weighting; second, a non-convex optimization model maximizing the Omega ratio is formulated and solved via a novel particle swarm optimization algorithm featuring operator self-adaptation and constraint-projection repair mechanisms. Empirical analysis on 421 constituents of the STOXX Europe 600 index demonstrates that the proposed ESG-enhanced strategy significantly outperforms both conventional non-ESG approaches and the market-capitalization-weighted benchmark, delivering higher risk-adjusted returns while advancing sustainable investment objectives.
0 citationsRead paper
A Structural Matrix Autoregressive Model for the Joint Dynamics of Volume, Volatility, and Returns
This study investigates the dynamic interactions and structural drivers among asset returns, realized volatility, and trading volume in high-dimensional financial data. To this end, the authors propose a Structured Matrix Autoregressive (SMAR) model that integrates identification constraints derived from the mixture-of-distributions hypothesis and the efficient market hypothesis, thereby preserving parameter parsimony while effectively capturing dynamic spillovers and cross-sectional dependencies. Empirical results reveal that volatility is a primary driver of trading volume; short-term volume dynamics are dominated by idiosyncratic shocks, whereas over 50% of long-term variation stems from cross-asset spillovers. Moreover, information-driven trading significantly intensifies on FOMC announcement days and exhibits rapid mean reversion. The proposed framework offers an identifiable and interpretable approach to modeling dynamics in high-dimensional financial systems.
0 citationsRead paper
Recent publications
Latest Papers
Leveraging contextual events on structure-aware next activity prediction
本文通过引入基于实例图的方法并结合上下文信息，利用图神经网络改进了流程中下一个活动的预测性能。
0 citationsRead paper
Theia: Large-Scale Multimodal Captioning and Automated Validation of the Incidents1M Dataset for Data-Free Distillation
This study addresses the scarcity of high-quality, large-scale multimodal datasets with semantically aligned image-text pairs in disaster response, which hinders data-free knowledge distillation for vision-language models. Building upon the purely visual Incidents1M dataset, the authors leverage the Qwen3.5 series of large language models to generate image captions and introduce an innovative image-blinded LLM-as-a-Judge automated verification mechanism. This approach simulates the modality gap faced by student models during distillation, effectively identifying and suppressing annotation inconsistencies and false positives inherent in human labeling. The method produces high-fidelity descriptions for 100,000 images, achieving a semantic consistency score of 78.65 on 173,179 annotated pairs and a verification accuracy of 77.6%, substantially enhancing the semantic reliability of data-free distillation.
0 citationsRead paper
A forgery attack on the Block.co blockchain-based digital credential certification system
This study addresses a critical vulnerability in existing blockchain-based academic credential verification systems, such as Block.co, which lack a trustworthy mechanism to securely bind an educational institution’s real-world identity to its digital counterpart. The authors present the first practical forgery attack against such systems, exploiting the disconnect between blockchain immutability and identity validation logic to generate fraudulent credentials that are erroneously validated as legitimate. This work exposes a fundamental flaw in decentralized credentialing models—namely, the insecure identity-binding process—and demonstrates that this weakness introduces systemic security risks across similar platforms. The proposed attack methodology is generalizable, highlighting a pervasive threat that undermines the integrity of blockchain-issued digital credentials and offering crucial insights for the secure design of future systems.
0 citationsRead paper
A Two-Stage Decision Support System for Sustainability-Aware Long Short Portfolio Optimization
This study addresses the challenge of constructing long–short portfolios that simultaneously achieve sustainability and superior risk-adjusted returns by integrating environmental, social, and governance (ESG) factors in a market-regime-adaptive manner. The authors propose a two-stage decision framework: first, assets are selected for long and short positions using a hybrid approach combining TODIMSort multi-criteria classification with MEREC-based weighting; second, a non-convex optimization model maximizing the Omega ratio is formulated and solved via a novel particle swarm optimization algorithm featuring operator self-adaptation and constraint-projection repair mechanisms. Empirical analysis on 421 constituents of the STOXX Europe 600 index demonstrates that the proposed ESG-enhanced strategy significantly outperforms both conventional non-ESG approaches and the market-capitalization-weighted benchmark, delivering higher risk-adjusted returns while advancing sustainable investment objectives.
0 citationsRead paper
A Structural Matrix Autoregressive Model for the Joint Dynamics of Volume, Volatility, and Returns
This study investigates the dynamic interactions and structural drivers among asset returns, realized volatility, and trading volume in high-dimensional financial data. To this end, the authors propose a Structured Matrix Autoregressive (SMAR) model that integrates identification constraints derived from the mixture-of-distributions hypothesis and the efficient market hypothesis, thereby preserving parameter parsimony while effectively capturing dynamic spillovers and cross-sectional dependencies. Empirical results reveal that volatility is a primary driver of trading volume; short-term volume dynamics are dominated by idiosyncratic shocks, whereas over 50% of long-term variation stems from cross-asset spillovers. Moreover, information-driven trading significantly intensifies on FOMC announcement days and exhibits rapid mean reversion. The proposed framework offers an identifiable and interpretable approach to modeling dynamics in high-dimensional financial systems.
0 citationsRead paper