Soroush Vosoughi awarded Google Research Scholar Grant

Assistant Professor Soroush Vosoughi has received the Google Research Scholar Award, for his work in developing culture-aware language models. The Google research program supports early-career faculty whose state-of-the-art research will impact how future generations use technology.

Vosoughi's work enables natural language processing (NLP) models to recognize and learn cultural  signatures in text. Cultural background can affect writing style. People from countries speaking the same language may use different expressions, such as vocabulary and idioms when describing or responding to the same event. Mixed writing styles in a dataset can become noisy for NLP models and negatively affect their training efficiency. Vosoughi aims to build accurate cultural background modeling based on written text and improve the performance of NLP models on other tasks by injecting cultural information into text representations.