Software Engineering Institute

Recent advancements in text analytics demonstrate significant gains in many natural language processing tasks, taking advantage of newer neural network architectures and transfer learning. In this article, we present findings from a literature review and exploratory study which investigated a specific application of text analytics for insider risk analysis. Results from our literature review find that recent advancements in text analytics greatly augment capabilities to exploit unstructured text commonly collected for insider risk analysis, offering new abilities to extract and generate intelligence to further support, standardize, and automate workflows. Results from our exploratory study suggest that the curation of keyword lists for insider threat detection can be augmented by automated text analytics capabilities, finding evidence that the manual process of managing threat keyword lists can be augmented with automated text analytics approaches. As a takeaway, the insider risk community should investigate further applications of text analytics to support and automate insider risk analysis workflows.