AI empowered senstive information detection and anonymisation in PDF files

In the digital age, privacy protection is critical concern in data exchange and storage. PDF (Portable Document Format) is a widely used file format that often carriers of sensitive information. To enhance privacy protection, effective anonymization of sensitive data within PDFs before public release is crucial. Using advanced OCR (Optical Character Recognition) and machine learning techniques, this paper presents a novel PDF anonymization framework capable of accurately identifying, detecting, and anonymizing sensitive information across various formats (numeric, categorical, textual). To ensure robust anonymization, we employ a combination of secure techniques, including k-Anonymity, l-Diversity, and t-Closeness. These techniques effectively mask sensitive information while preserving data utility. Experimental results validate the efficacy of our proposed solution.