Group-based Password Characteristics Analysis

Abstract

In this article, we analyze password characteristics from the perspective of user groups in different countries and web-based services. We collect a dataset from the Chinese railway website www.12306.cn. which contains data from four provinces, Hubei, Zhejiang, Inner Mongolia and Xinjiang. Additionally, we select datasets from two English based Internet applications, Faithwrit-er and Facebook. We analyze these six datasets based on several common indicators, including popular passwords, password structure and letter distribution. The analysis results show that there are remarkable differences in different user groups. The experiments show that geographical factors (embodied in the native language) and types of website services play a significant role in password creation. We further evaluate the security of these passwords by employing two state-of-the-art password cracking techniques. The attack results show that datasets of different provinces and different types of website services have different password strength. To the best of our knowledge, this is the first time passwords are analyzed based on different user groups.