Selected work

Representative Papers

Financial Dynamics and Interconnected Risk of Liquid Restaking

Mar 23, 2026arXiv.org

This study addresses the complex systemic risks introduced by liquid restaking, which, while enhancing yield, demands empirical analysis of its return drivers and risk transmission mechanisms. Focusing on the Renzo protocol, the work integrates DeFi asset flow mapping, OLS regression, Granger causality tests, and random forest feature importance to identify EigenLayer’s total value locked, liquid restaking token yields, and multi-chain expansion as key predictors of returns—revealing the latter’s dual-edged nature. By constructing a cross-protocol risk transmission framework and conducting stress tests, the analysis finds that current bridge-related risks do not yet pose a systemic threat; however, the growing interconnectivity warrants ongoing vigilance.

1 citationsRead paper

New Perspectives on Semiring Applications to Dynamic Programming

Dec 03, 2025

Counting minimum-cost solutions to NP-hard combinatorial optimization problems—such as Connected Dominating Set and Constraint Satisfaction—is computationally challenging due to the interplay between cost minimization and solution enumeration. Method: We propose a unified dynamic programming framework grounded in semiring algebra, centered on a novel Δ-product operation that relaxes the idempotence requirement of classical semirings, enabling efficient counting of minimum-cost solutions over non-idempotent semirings. The framework is parameterized by treewidth and clique-width to ensure tractability on structured inputs. Contribution/Results: We establish fixed-parameter tractable (FPT) enumeration for minimum-cost solutions under bounded treewidth or clique-width, proving polynomial-time solvability of the counting problem in these parameters. Our approach significantly extends classical DP’s expressiveness, unifying cost optimization and solution counting within a single algebraic model—thereby enabling rigorous analysis and efficient computation for previously intractable enumeration tasks.

1 citationsRead paper

RelShap: Relationally Consistent Shapley Explanations

Aug 11, 2026

This work addresses a critical limitation of traditional Shapley value methods, which assume feature independence and thereby ignore structural constraints inherent in relational data—such as functional dependencies—leading to the generation of invalid feature coalitions and distorted explanations. To overcome this, the authors propose RelShap, a novel framework that incorporates relational integrity into Shapley-based explanations by restricting coalition formation to only those subsets that satisfy data validity constraints. Leveraging functional dependencies, RelShap constructs equivalence classes of coalitions, preserving explanation consistency while substantially reducing computational complexity. The framework is compatible with mainstream estimators such as Kernel SHAP and Monte Carlo SHAP. Empirical evaluations across multiple datasets and models demonstrate that RelShap more accurately captures the true data-generating mechanism and outperforms existing approaches, including Conditional SHAP and ManifoldSHAP.

0 citationsRead paper

Incidental Visualizations: Augmented Reality as a Medium for Contextual Information

Aug 07, 2026

This study addresses the challenge of efficiently conveying contextual information without disrupting users’ primary tasks. It proposes a novel paradigm termed “incidental visualizations,” which leverages augmented reality (AR) to transiently and spontaneously embed information into the user’s environment. Through user experiments involving logical puzzle tasks such as Sudoku and Connect Four, the approach is compared against persistent and periodic visualization strategies. Results demonstrate that incidental visualizations achieve information comprehension accuracy comparable to persistent displays while significantly reducing interference with the main task. This work presents the first systematic definition and empirical validation of this low-intrusion, contextually relevant AR information delivery method, offering a new direction for the design of adaptive AR interfaces.

0 citationsRead paper

A Unified Feature Model for Microservice Identification and Refactoring

Jul 31, 2026

This study addresses the fragmentation and lack of interoperability among existing microservice identification approaches by proposing the first unified framework based on a feature model. The framework systematically integrates diverse identification and refactoring methods through a systematic literature mapping, constructing a feature model that captures the primary variation points across current techniques. By analyzing concrete tool architectures instantiated within this model, the work reveals both the complementary nature of existing tools and the comprehensiveness of the overall design space. Results demonstrate that while individual tools cover only limited subsets of the variation points, the collective set of available tools nearly spans the entire spectrum defined by the feature model. This provides a solid theoretical foundation and practical guidance for future tool development and integration in microservice identification.

0 citationsRead paper

Recent publications

Latest Papers

RelShap: Relationally Consistent Shapley Explanations

Aug 11, 2026

This work addresses a critical limitation of traditional Shapley value methods, which assume feature independence and thereby ignore structural constraints inherent in relational data—such as functional dependencies—leading to the generation of invalid feature coalitions and distorted explanations. To overcome this, the authors propose RelShap, a novel framework that incorporates relational integrity into Shapley-based explanations by restricting coalition formation to only those subsets that satisfy data validity constraints. Leveraging functional dependencies, RelShap constructs equivalence classes of coalitions, preserving explanation consistency while substantially reducing computational complexity. The framework is compatible with mainstream estimators such as Kernel SHAP and Monte Carlo SHAP. Empirical evaluations across multiple datasets and models demonstrate that RelShap more accurately captures the true data-generating mechanism and outperforms existing approaches, including Conditional SHAP and ManifoldSHAP.

0 citationsRead paper

Incidental Visualizations: Augmented Reality as a Medium for Contextual Information

Aug 07, 2026

This study addresses the challenge of efficiently conveying contextual information without disrupting users’ primary tasks. It proposes a novel paradigm termed “incidental visualizations,” which leverages augmented reality (AR) to transiently and spontaneously embed information into the user’s environment. Through user experiments involving logical puzzle tasks such as Sudoku and Connect Four, the approach is compared against persistent and periodic visualization strategies. Results demonstrate that incidental visualizations achieve information comprehension accuracy comparable to persistent displays while significantly reducing interference with the main task. This work presents the first systematic definition and empirical validation of this low-intrusion, contextually relevant AR information delivery method, offering a new direction for the design of adaptive AR interfaces.

0 citationsRead paper

A Unified Feature Model for Microservice Identification and Refactoring

Jul 31, 2026

This study addresses the fragmentation and lack of interoperability among existing microservice identification approaches by proposing the first unified framework based on a feature model. The framework systematically integrates diverse identification and refactoring methods through a systematic literature mapping, constructing a feature model that captures the primary variation points across current techniques. By analyzing concrete tool architectures instantiated within this model, the work reveals both the complementary nature of existing tools and the comprehensiveness of the overall design space. Results demonstrate that while individual tools cover only limited subsets of the variation points, the collective set of available tools nearly spans the entire spectrum defined by the feature model. This provides a solid theoretical foundation and practical guidance for future tool development and integration in microservice identification.

0 citationsRead paper

There Will Be Spam: Characterizing State-Invariant Transactions and Speculative MEV

Jul 27, 2026

This study addresses the prevalence of state-invariant transactions—those that do not alter the blockchain’s final state—which waste resources and undermine decentralization, particularly on Layer-2 networks. For the first time, such transactions are systematically identified as on-chain spam through large-scale on-chain data analysis, transaction replay validation, and MEV strategy classification across Ethereum, Optimism, and Base. The findings reveal that 24% of transactions on Optimism and 37% on Base are state-invariant, with speculative MEV accounting for 57%–68% of these. On Ethereum, address poisoning constitutes 53% of non-reverted transactions. Moreover, the research demonstrates that the profitability of speculative MEV is significantly overestimated and uncovers widespread malicious activity, challenging prevailing assumptions about the origins of on-chain spam.

0 citationsRead paper

Identity-Bound Academic Credentials on Blockchain: On-Chain Issuer Accreditation with ERC-3643 and OnchainID

Jul 17, 2026

This work addresses critical limitations in existing academic credential systems—such as institutional silos, inefficient verification, rampant forgery, lack of trusted issuer authentication, and inadequate support for credential updates or revocation—by proposing a blockchain-based academic credential registry. It pioneers the adaptation of the financial-oriented ERC-3643 standard to non-transferable academic credentials, integrating OnchainID (ERC-734/735) and the T-REX framework to enable on-chain issuer authentication, student identity binding, third-party verifiability, and lifecycle management including updates and revocations, while keeping sensitive data off-chain. The system supports signed claims, wallet-less verification, and off-chain storage, providing a full-lifecycle reference implementation. Comprehensive quantitative evaluation demonstrates its performance in gas consumption, scalability, latency, and security, clearly delineating the safe applicability boundaries of ERC-3643 in academic contexts.

0 citationsRead paper