How text mining can be used to detect fraud

Many fraud examiners may be more familiar with number-crunching than parsing words; but, as a use-case shows below, high-volume text analysis can be of significant assistance in identifying deceit. Lacey Keller of MK Analytics and Erik Halvorson of the US Department of Energy set out methodology for text extraction from diverse sources, as well as analysing the content.

Lacey Keller, an expert in data mining and analytics, is co-founder of MK Analytics (www.mk-analytics.com). Erik Halvorson is a special agent focused on anti-fraud programme management for the US Department of Energy's new funding expenditures within the Office of Investigations (https://nij.ojp.gov/bio/erik-halvorson) and is a small business owner. If you have questions about this topic or would like to discuss it further, the authors can be reached at: lkeller@mk-analytics.com or erik.halvorson@archimedeseval.com

Many fraud examiners may be more familiar with number-crunching than parsing words; but, as a use-case shows below, high-volume text analysis can be of significant assistance in identifying deceit. Lacey Keller of MK Analytics and Erik Halvorson of the US Department of Energy set out methodology for text extraction from diverse sources, as well as analysing the content.