Abstract: Chinas Data Security Law regards classification and grading as the basic system for data protection. It stipulates that public security organs should be responsible for the security of data collected and generated in the course of their work. However, the current public security industry standards lack a basis for the classification of public security data security. This study measures the privacy value of personal data within the public security system and provides a quantitative basis for data classification. A privacy text database is constructed by screening four types of privacy texts in the public security field. Semantic elements such as public security data items, core verbs, and degree modifiers, are extracted to establish a public security privacy semantic lexicon. Based on the sensitivity, semantic strength, and textual intensity of privacy vocabulary, a privacy measurement model is constructed to measure the privacy value of personal data items in public security. According to the measurement results, public security personal data are classified into four levels: (1) criminal and economic investigation data, (2) personal information data, (3)basic personal data and public security management data, and(4)traffic management data.