Developing a free DLP system for email data leakage

Data leakage (or data loss) is a term used in the information security field to describe unwanted disclosures of information. According to DataLossDB site data breaches (leakages) in 2015 are increased dramatically. According to Risk Based Security 2016 Data Breach Report analysis, while the percentage of total breaches impact 89.5% electronic data, it is 99.6% in 2016. In this thesis, a lite e-mail data leakage prevention (DLP) solution using open source projects for e-mail data leakage has been proposed. A dirty filter script running on top of the EmailRelay server is developed and the system is tested under different loads. Different parsing architectures are tried and searched for the most efficient one. Also the proposed solution is compared with another known solution. For this purpose the community version of MyDLP server is evaluated. The outgoing e-mail with binary attachment content is analysed with Apache TIKA framework. The e-mail content which is converted to textual data, is searched for data leakage patterns using regular expressions. In most of the studies the scanned image documents are not analysed for leakage. In this study open source tools are used to convert the scanned image document to textual data; because most of the documents in the Internet are scanned confidential documents. Finally, the leaked e-mail is indexed with Elasticsearch for further contextual analysis.