NTUA-SLP at SemEval-2018 Task 2: Predicting Emojis using RNNs with Context-aware Attention

TL;DR

This paper introduces a deep learning model using RNNs with context-aware attention for emoji prediction in tweets, achieving high accuracy without handcrafted features.

Contribution

The novel approach combines LSTM with a context-aware attention mechanism and pretrained embeddings for effective emoji prediction in multilingual tweets.

Findings

Ranked 2nd out of 48 teams in SemEval-2018 Task 2

Achieved strong performance without handcrafted features

Utilized large-scale pretrained embeddings for improved accuracy

Abstract

In this paper we present a deep-learning model that competed at SemEval-2018 Task 2 "Multilingual Emoji Prediction". We participated in subtask A, in which we are called to predict the most likely associated emoji in English tweets. The proposed architecture relies on a Long Short-Term Memory network, augmented with an attention mechanism, that conditions the weight of each word, on a "context vector" which is taken as the aggregation of a tweet's meaning. Moreover, we initialize the embedding layer of our model, with word2vec word embeddings, pretrained on a dataset of 550 million English tweets. Finally, our model does not rely on hand-crafted features or lexicons and is trained end-to-end with back-propagation. We ranked 2nd out of 48 teams.