Discovering Personality Traits from Twitter

Gou, L., Zhou, M. X., & Yang, H. (2013). KnowMe and ShareMe: Understanding automatically discovered personality traits from social media and user sharing preferences. San Jose, CA: IBM Research Division. Retrieved from http://domino .research.ibm.com/library/cyberdig.nsf/papers/9A604B598317CE2785257C55005A25A6/$File/rj10516.pdf

In this paper, the authors present findings from a two-part study that: (a) “examined the effectiveness of automatically deriving three types of personality traits from Twitter” (i.e., Big Five personality traits, basic human values, and fundamental needs), and (b) “investigated users’ opinions of using and sharing these traits” (p. 1). Data were collected from 256 participants from the authors’ company (IBM) with 224 complete responses. The authors first identified participants’ personality types from analysis of participants’ tweets (collected through an experimental system they built called KnowMe, which automatically derives personality traits), as well as their psychometric test scores and perceptions (collected via a two-part online survey). The authors then designed and administered a separate questionnaire that assessed participants’ privacy preferences and the factors that influence them in four areas: attributes, actors, context, and transmission principles. The results of correlational and other analyses identified user preferences as well as perceived risks/benefits of sharing their information. The authors provided recommendations for making the system more effective, noting user desire for system transparency in terms of system usage and function, the need for a prescriptive system that has clear capabilities and limitations, and the importance of protecting users’ privacy. The authors provided also practical recommendations regarding improving the automated personality discovery capability of the system.  

‍