Toward the Development of a Psycholinguistic-based Measure of Insider Threat Risk Focusing on Core Word Categories Used in Social Media

The development of a psycholinguistic measure of insider threat risk faces many challenges. Prior research suggests that psycholinguistic analysis has potential for identifying individuals who are high risks for insider/criminal activity, but testing and validation of proposed methods is hampered by lack of available data. For example, analytic tools can discriminate text samples authored by known criminals from e-mail text produced in an organization, but a robust risk assessment tool must be able to handle heterogeneous text sources that come from a diverse population with wide differences in word use across age and other demographic variables. The current study seeks to advance the state of psycholinguistic insider threat research by among other things, augmenting word analysis dictionaries used in current practice and accommodating linguistic characteristics of social media, focusing on a core set of differentially-weighted word categories that represent an intersection of critical personality traits related to behaviors of concern.