摘要： 基于用户所特有的击键节奏特征,提出了一种判别某一击键序列是否为某特定用户的行为的方案。该方案利用非参数方法中的Spearman 相关分析来确定计算 cityblock 距离所需的权值,进而利用预存样本与新输入样本之间的距离来判定用户的身份。所以此方案不受各种可能击键的概率分布限制。一个基于击键特征的用户身份校验系统已经使用 Java 语言成功地实现。

关键词: 身份校验, 击键特征, 非参数方法

Abstract: Based on characteristic typing style of an arbitrary user, a scheme to distinguish whether a keystroke series is made by a specific user or not is proposed. The cityblock distance between samples stored in advance and the new input sample is the criterion of identity verification. This scheme uses a nonparametric method, that is, Spearman correlation analysis, to calculate the weights for computing the cityblock distance. So the variety of possible probability distributions of keystrokes is not a regulation. A user identity verification system also has been implemented successfully with Java.

中图分类号: 

• TP309