A Survey of Data Leakage Detection and Prevention Techniques: Fingerprinting, Access Control, Machine Learning, and Self-Protective Data Objects

Abstract

Data leakage is a major information-security problem because sensitive information can be exposed during storage, use, transfer, or sharing with authorised third parties. The supplied research material reviews data leakage detection and prevention (DLD/DLP) mechanisms spanning data allocation with fake objects, watermarking, k-anonymity, access control, behavioural analysis, runtime protection, data lineage, encryption, redaction, steganography, and content-analysis techniques. This survey organises these approaches according to the data life cycle and the degree of automation. Particular attention is given to insider threats, because authorised users can misuse legitimate credentials and are difficult to distinguish from normal users. The survey further examines the evolution from static rule-based controls toward adaptive and self-protective data objects. The supplied work identifies a research gap in dynamically configurable protection and motivates a combined architecture using policy enforcement, fingerprinting, regular expressions, probabilistic prediction, and runtime scrutiny. The review indicates that no single technique is sufficient for all leakage modes; effective protection requires layered controls that combine prevention, detection, attribution, and rapid reporting.