Institution profile

University of Salerno

Academic institutioneurope · it
Official website
Research library146linked papers
Opportunities0open roles
Selected work

Representative Papers

Smart Contract Vulnerabilities, Tools, and Benchmarks: An Updated Systematic Literature Review

Dec 02, 2024arXiv.org

Ethereum smart contracts face severe security challenges due to immutability and transparency, yet existing vulnerability detection tools lack systematic evaluation of effectiveness. To address this gap, we conduct a large-scale, systematic literature review (SLR), rigorously selecting and analyzing 222 high-quality papers from an initial pool of 3,380 studies. We propose the first hierarchical taxonomy covering 14 vulnerability categories and 192 distinct vulnerability types. Further, we introduce a novel “tool–vulnerability–benchmark” ternary mapping framework, unifying 219 detection tools and 133 evaluation benchmarks for the first time. Our contributions include: (1) a standardized vulnerability classification schema; (2) a comprehensive functional matrix of all surveyed tools; (3) a fine-grained tool–vulnerability matching atlas; and (4) a reusable, harmonized benchmark suite. This work establishes an authoritative, extensible knowledge infrastructure for smart contract security research, substantially enhancing comparability, reproducibility, and methodological rigor in vulnerability detection studies.

2 citationsRead paper

Predictive Models for Chronic Heart Failure

Apr 07, 2025

To address the clinical challenge of early risk identification in chronic heart failure (HF) patients and the suboptimal sensitivity of existing predictive models, this study proposes a dual-path decoupled stacked ensemble framework integrating clinical and echocardiographic features. Distinct base learners model each heterogeneous feature modality separately, and a meta-learner fuses their outputs into a calibrated risk score—prioritizing high sensitivity as required in clinical screening. Evaluated on real-world data, the model achieves 95% sensitivity and 84% accuracy, enabling precise high-risk patient identification in the PrediHealth remote monitoring program. The core contributions are: (1) the first dual-path feature decoupling mechanism for HF risk prediction, preserving modality-specific signal integrity; and (2) a clinically optimized stacked architecture that enhances recall while maintaining interpretability. This yields a robust, explainable, and high-recall risk stratification tool tailored to clinical deployment requirements.

1 citationsRead paper

Augmenting Anonymized Data with AI: Exploring the Feasibility and Limitations of Large Language Models in Data Enrichment

Apr 03, 2025

Privacy-preserving techniques such as k-anonymity often induce data sparsity, information loss, and degraded downstream modeling performance. Method: This paper proposes a large language model (LLM)-based approach for anonymized data augmentation, centered on a privacy-constraint-aware, customized prompting strategy that rigorously enforces k-anonymity, l-diversity, and t-closeness requirements. The method integrates structured prompt templates, the pyCanon privacy verification framework, and real-world datasets to enable verifiable, privacy-compliant data expansion. Contribution/Results: We present the first systematic evaluation of LLMs’ capability to generate privacy-compliant synthetic data. Experimental results demonstrate that the augmented data significantly enhances feature richness and improves downstream prediction accuracy by an average of +8.2%, without violating anonymity constraints—establishing a novel paradigm for jointly optimizing privacy protection and model utility.

1 citationsRead paper
Recent publications

Latest Papers