We develop a software system Text Scanner for Emotional Distress (TSED) for helping to detect email messages which are suspicious of coming from people under strong emotional distress. It has been confirmed by multiple studies that terrorist attackers have experienced a substantial emotional distress at some points before committing a terrorist attack. Therefore, if an individual in emotional distress can be detected on the basis of email texts, some preventive measures can be taken. The proposed detection machinery is based on extraction and classification of emotional profiles from emails. An emotional profile is a formal representation of a sequence of emotional states through a textual discourse where communicative actions are attached to these emotional states. The issues of extraction of emotional profiles from text and reasoning about it are discussed and illustrated. We then develop an inductive machine learning and reasoning framework to relate an emotional profile to the class "Emotional distress" or "No emotional distress", given a training dataset where the class is assigned by an expert. TSED's machine learning is evaluated using the database of structured customer complaints.

Date Published: 18 April 2006
PDF: 11 pages
Proc. SPIE 6241, Data Mining, Intrusion Detection, Information Assurance, and Data Networks Security 2006, 62410G (18 April 2006); doi: 10.1117/12.666645

Published in SPIE Proceedings Vol. 6241:
Data Mining, Intrusion Detection, Information Assurance, and Data Networks Security 2006
Belur V. Dasarathy, Editor(s)