Blockchain Papers

Santosh S Doifode, Anand Singh Rajawat

No abstract is available for this record.

S. Lakshmi

Persuasive textual narratives, bogus visual evidence, disreputable update patterns and absent accountability systems are being increasingly used to deceive backers in fraudulent crowdfunding campaigns. Current fraud detection techniques are primarily based on static information, on text-only indicators, or on very shallow fusion of multimodal information, and they are not able to detect deceptive information that evolves over time or is inconsistent across different modalities. This study presents a Temporal Cross-Modal Trust Intelligence Framework to mitigate reward-based crowdfunding fraud that is explainable. The framework combines Hidden Method-of-Moments Markov modelling for latent temporal behaviour analysis, Polynomial Expansion Canonical Correlation Analysis for nonlinear text–image consistency evaluation and a Frequency-Gated GRU classifier to distinguish subtle drift in behaviour from sudden suspicious behaviour anomalies. Local Outlier Factor-based risk refinement is also added to detect rare and locally abnormal fraud patterns, and a blockchain-auditable layer ensures prediction outcomes are transparent and tamper-proof, enhancing the decision-making process. Experimental results on a multimodal crowdfunding dataset created from the Kickstarter platform show that the proposed model achieves better accuracy, recall, F1-score, ROC-AUC, PR-AUC, and calibration reliability than conventional multimodal, transformer-based, and recurrent neural network models and classical machine learning. The results validate the proposed solution, which is built on the four temporal dynamics, cross-modal consistency, anomaly refinement and auditability, to be a comprehensive and interpretable solution for detecting early crowdfunding fraud.

Bharath M B, Latha N R

Web security has become a critical domain as modern applications increasingly rely on dynamic user-generated content, making them highly vulnerable to Cross-Site Scripting (XSS) attacks. Traditional detection systems struggle to cope with evolving payload patterns, limited generalisation across institutions, and strict privacy restrictions that prevent sharing of sensitive request logs. To address these challenges, this work proposes a Zero-Knowledge Federated Sequence Learning (ZK-FSL) framework that enables collaborative XSS detection without exposing raw data or intermediate gradients. The model integrates attention-based deep sequence learning with zero-knowledge proof validation, ensuring both strong predictive capability and verifiable trust among participating clients. Experimental evaluation demonstrates that ZK-FSL achieves superior performance compared to centralised and federated baselines, reaching 96.3% accuracy , 96.7% precision , 95.9% recall , 96.3% F1-score , and an AUC of 0.98 . These results confirm that the proposed framework effectively enhances privacy-preserving threat detection while maintaining high robustness against diverse and sophisticated XSS attack patterns.