中文文本中的个人信息自动化检测框架研究  

AUTOMATED DETECTION FRAMEWORK FOR PERSONAL INFORMATION IN CHINESE TEXT

作　　者：贾昊阳 盛毅敏 阮雯强 韩伟力[1] Jia Haoyang;Sheng Yimin;Ruan Wenqiang;Han Weili(Software School,Fudan University,Shanghai 201203,China)

机构地区：[1]复旦大学软件学院,上海201203

出　　处：《计算机应用与软件》2022年第2期307-314,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(U1836207);上海市创新行动计划项目(18511103600)。

摘　　要：为帮助数据持有者规避法律风险,有必要对数据集中的个人信息做检测和统计。然而当前尚缺有效工具支持检测中文数据集中的个人信息。为应对上述问题,根据法律文献整理出需要检测的个人信息类别,提出综合了模式匹配与自然语言处理技术的个人信息自动化检测框架,对中文文本中的个人信息进行检测。同时,提出一种识别家庭住址的方法,解决地址格式多样化的问题。以在线社交网络平台上的用户发布数据以及大规模的公开文本数据为例,展示该框架的可行性和有效性。To help data holders avoid legal risks,it is necessary to detect personal information in datasets.However,there is currently no effective tool to support the detection of personal information in Chinese datasets.To address above issues,we specified the categories of personal information that need to be detected by investigating privacy protection laws,and we proposed an automated detection framework,which integrated pattern matching algorithms and natural language processing techniques to detect the specified personal information in Chinese text.We also proposed a home address recognition method that overcame the complexity of address formats.The feasibility and effectiveness of this framework are demonstrated by detecting user-generated data on online social networks and large-scale textual data available on the Internet.

关 键 词：个人信息 隐私检测 自动化 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]