Yejin studies computational models for human language processing, with a focus on contextual and nonliteral language understanding. Through data-driven analysis of writing style, Yejin built computer systems for interpreting the communicative intent and the intellectual identity of the authors; for example, detecting deceptive reviews in online communities and predicting literary success of novels. In addition, connecting language understanding with computer vision, Yejin studied data-driven models for automatically captioning online photographs and for inferring the natural conceptualization of the object names in the world.

Her work has been features in numerous media outlets, including the New York Times, WNBC News, IEEE Spectrum Podcast, NPR, and CBS Radio and was awarded the 2013 David Marr Prize.