A Computational Study on Emotional Responses via Amodal Propagation: Dimensional Vs. Discrete Emotions

EasyChair Preprint 3690

Abstract

We computationally emulated a link between symbolic and emotional representations of words using computational models and predictive models. We studied dimensional and discrete emotions using two different predictive models: linear regressions and neural networks. More than 13000 words were used to train the models and then they were tested in more than 4000 words. While important differences were observed between linear regressions and neural networks in dimensional emotions, no differences were observed in discrete emotions.

Keyphrases: emotion, linear regression, neural networks

@booklet{EasyChair:3690,
  author    = {José Á. Martínez-Huertas and Guillermo Jorge-Botana and Ricardo Olmos and Alejandro Martínez-Mingo},
  title     = {A Computational Study on Emotional Responses via Amodal Propagation: Dimensional Vs. Discrete Emotions},
  howpublished = {EasyChair Preprint 3690},
  year      = {EasyChair, 2020}}