In this paper, we present an alerting system based on an efficient classification model for detecting suicidal people using natural language processing and data mining techniques. The model uses linguistic features which are derived from an analysis of handwritten and electronic messages/notes. The model was trained and validated with fully anonymised real data provided by the Cyber Crime Division of Greek Police as well as available suicidal notes from social media. The alerting system is intended as a prevention, management tool for automatic detection of suicidal intentions.