Financial Discussion Boards Irregularities Detection System (FDBs-IDS) using Information Extraction

Owda, M, Crockett, K and Lee, P (2017) Financial Discussion Boards Irregularities Detection System (FDBs-IDS) using Information Extraction. In: Intelligent Systems Conference 2017, 07 September 2017 - 08 September 2017, London, UK. (In Press)

Preview

Download (692kB) | Preview

Abstract

The current growth and the use technology in global stock markets has created unprecedented opportunities for the individuals and businesses to access capital and grow and diversify their portfolios. Individuals, nowadays can decide to invest and act in few minutes if not in few seconds. This growth has led to a corresponding growth in the amount of fraud and misconduct seen in the stock markets through the use of technology. The internet is often used as a real time platform for illegal financial activity such as illegal activities on Financial Discussion Boards (FDBs). Managing and monitoring FDBs in real time is a complex and time consuming task; given the volume of data produced and the fact that some of the data is unstructured. This paper presents a novel Financial Discussion Boards Irregularities Detection System (FDBs-IDS) for FDBs which can highlight irregularities or potentially unlawful practices on FDBs. For example comments that might suggest a pump and dump activity is happening. The proposed system extracts information from FDBs, where templates hosting scenarios of known illegal activities are used to detect any potential misdemeanors. Analysis conducted on a single day trading, found that of the 3000 comments extracted from FDBs, 0.2% of these comments were deemed suspicious and required further investigation of a discussion board moderator. The man-power required to perform this task manually over the course of a year could be excessive and unaffordable. This research highlights the importance and the need of an automated crime detection system on FDBs such as FDBs-IDS which could be used and thus tackle potential criminal activities on the internet.