Sentylic at IEST 2018: Gated Recurrent Neural Network and Capsule Network Based Approach for Implicit Emotion Detection

TL;DR

This paper introduces a neural network model combining Gated Recurrent Units and Capsule Networks to detect implicit emotions in tweets, leveraging pre-trained embeddings to improve understanding of context without explicit emotion terms.

Contribution

The paper presents a novel combination of GRU and Capsule Networks for implicit emotion detection, achieving competitive performance on the WASSA 2018 shared task.

Findings

Achieved a macro-F1 score of 0.692

Effectively captures implicit emotional cues in tweets

Demonstrates the utility of capsule networks in NLP tasks

Abstract

In this paper, we present the system we have used for the Implicit WASSA 2018 Implicit Emotion Shared Task. The task is to predict the emotion of a tweet of which the explicit mentions of emotion terms have been removed. The idea is to come up with a model which has the ability to implicitly identify the emotion expressed given the context words. We have used a Gated Recurrent Neural Network (GRU) and a Capsule Network based model for the task. Pre-trained word embeddings have been utilized to incorporate contextual knowledge about words into the model. GRU layer learns latent representations using the input word embeddings. Subsequent Capsule Network layer learns high-level features from that hidden representation. The proposed model managed to achieve a macro-F1 score of 0.692.