Improving Relevance Effectiveness in Data Leakage Detection Using Feature Selection

摘要

Data leakage is an uncontrolled or unauthorized transmission of classified information to the outside. Many software solutions were developed to provide data protection. However, none of them can provide absolute protection. The purpose of the research is to design and implement DATALEAK, a data leakage detection system based on information retrieval models and methods. In this paper, a feature selection based information retrieval model is proposed to improve relevance effectiveness of DATALEAK. The paper focuses on dimensionality reduction, where semantic matching of documents is performed in the reduced form of the vector space model.