Similarities: The Key Factors Influencing Cross-Site Password Guessing Performance

• DOI: 10.3390/electronics14050945
• Publication Date: 2025-02-28T15:46:46Z
• Authors: Zhijie Xie; Fan Shi; Min Zhang; Zhihong Rao; Yuxuan Zhou; Xiaoyu Ji

Password guessing is a crucial research direction in password security, considering vulnerabilities like reuse and data breaches. While has extensively explored intra-site guessing, the complexities of cross-site attacks, where attackers use leaked from one site to target another, remain less understood. This study investigates impact dataset feature similarity on performance, revealing that differences significantly influence success more than model variations. By analyzing eight datasets four methods, we identified key features affecting success, including general length distribution specific semantic PCFG grammar. Our reveals syntactic statistical patterns passwords, particularly features, are most effective for due their strong generalization across datasets. The Spearman correlation coefficient 0.754 between rate indicates significant positive correlation, unlike minimal (0.284). These findings highlight importance focusing robust improving techniques security strategies. Additionally, underscores selection suggests defenders can enhance by mitigating with commonly data.