GSCL MicroGrant awarded

Sabine Weber receives a GSCL MicroGrant for their project proposal "A Cross-Task Personalization Corpus for Emotion Classification, Humor Detection and Aggression Detection in Text". With this grant they plan to address the problem of low classifier performance in tasks that require subjective interpretation, such as in humor detection, aggression detection, and emotion classification. Previous studies have shown that tailoring models to individual annotators—known as personalization—can improve performance. However, in a system that covers several tasks, individual annotations for every task are necessary. In their project Sabine Weber will collect annotations from the same annotators across different tasks so that the personal annotation preferences of a person can be learned from one task and transferred to the other. Their work aims to facilitate research on the relationships among emotion, humor, and aggression in text and the development of better personalized models.