Computational Human Behavior
Reading people — emotion, cognition, and identity — from digital traces.
People leave rich traces in the language they write and the interactions they choose, and those traces reveal who they are and how they behave. I develop computational methods that read these signals: modeling individuals and populations from sparse, high-dimensional behavioral data, extracting emotion and cognition from user-generated content, identifying authors from their linguistic fingerprints, and tracing how reputation and influence take shape in social media. This connects user modeling and recommendation with information-theoretic representation — a principled way to characterize human behavior, and to catch the subtle shifts that mark a change in a person or a community.
Selected work: determining films’ evoked emotional experience from online reviews (Mokryn et al., 2020); the Movie Emotion Map (Cohen-Kalaf et al., 2022); cross-platform analysis of review usefulness (Mokryn, 2020); and inferring purchase intent from anonymous sessions (Mokryn et al., 2019).
References
2022
- MTAPMovie Emotion Map: An Interactive Tool for Exploring Movies According to their Emotional SignatureMultimedia Tools and Applications, 2022
2020
- IR J.Sharing Emotions: Determining Films’ Evoked Emotional Experience from their Online ReviewsInformation Retrieval Journal, 2020Best Research Paper Award, Israel Association for Information Systems (ILAIS), 2022
- OSNEMThe Opinions of a Few: A Cross-Platform Study Quantifying Usefulness of ReviewsOnline Social Networks and Media, 2020
2019
- ECRAWill This Session End with a Purchase? Inferring Current Purchase Intent of Anonymous VisitorsElectronic Commerce Research and Applications, 2019