Better Understanding Emotions in Texts: A Cognitive Hybrid Deep Learning Approach
Chang Liu, Keehyung Kim · Journal of the Association for Information Systems · 2020
With the widespread use of online platforms, people frequently share their thoughts and opinions via emotionally charged texts. For online service providers and researchers, these user-generated contents serve as valuable resource in analyzing individual’s view as well as providing personalized services. However, many of the existing methodologies and labeled dictionary datasets are lacking when there are multiple co-existing discrete emotions embedded in texts. In this research-in-progress study, we propose a new Cognitive Joint Attention Neural Network (CJANN) model inspired by the human’s cognitive process in reading texts. This model also incorporates three layers of attention modules to measure the level of emotion provision for eight discrete emotions. The proposed deep learning model outperforms other widely used models.