Messaging activity reconstruction with sentiment polarity identification

Andriotis, Panagiotis; Oikonomou, George

Abstract

© Springer International Publishing Switzerland 2015. Sentiment Analysis aims to extract information related to the emotional state of the person that produced a text document and also describe the sentiment polarity of the short or long message. This kind of information might be useful to a forensic analyst because it provides indications about the psychological state of the person under investigation at a given time. In this paper we use machine-learning algorithms to classify short texts (SMS), which could be found in the internal memory of a smartphone and extract the mood of the person that sent them. The basic goal of our method is to achieve low False Positive Rates. Moreover, we present two visualization schemes with the intention to provide the ability to digital forensic analysts to see graphical representations of the messaging activity of their suspects and therefore focus on specific areas of interest reducing their workload.

• Presentation Conference Type: Conference Paper (published)
• Acceptance Date: Aug 2, 2015
• Publication Date: Jan 1, 2015
• Journal: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
• Print ISSN: 0302-9743
• Publisher: Springer Verlag
• Peer Reviewed: Peer Reviewed
• Volume: 9190
• Pages: 475-486
• Book Title: Human Aspects of Information Security, Privacy, and Trust
• DOI: https://doi.org/10.1007/978-3-319-20376-8_42
• Keywords: smartphone, forensics, text-mining, short-text messages
• Public URL: https://uwe-repository.worktribe.com/output/840640
• Publisher URL: http://dx.doi.org/10.1007/978-3-319-20376-8_42
• Additional Information: Title of Conference or Conference Proceedings : Third International Conference, HAS 2015, Held as Part of HCI International 2015