# BB_twtr at SemEval-2017 Task 4: Twitter Sentiment Analysis with CNNs and LSTMs

## TL;DR

This paper presents a Twitter sentiment analysis system using CNNs and LSTMs trained on large unlabeled data, achieving top performance in SemEval-2017 by leveraging pre-trained embeddings and ensemble methods.

## Contribution

It introduces a novel combination of pre-trained embeddings, fine-tuning with distant supervision, and ensemble techniques for improved Twitter sentiment classification.

## Key findings

- Achieved first place in all five English subtasks at SemEval-2017.
- Utilized large unlabeled data for effective embedding pre-training.
- Ensemble of CNNs and LSTMs outperformed individual models.

## Abstract

In this paper we describe our attempt at producing a state-of-the-art Twitter sentiment classifier using Convolutional Neural Networks (CNNs) and Long Short Term Memory (LSTMs) networks. Our system leverages a large amount of unlabeled data to pre-train word embeddings. We then use a subset of the unlabeled data to fine tune the embeddings using distant supervision. The final CNNs and LSTMs are trained on the SemEval-2017 Twitter dataset where the embeddings are fined tuned again. To boost performances we ensemble several CNNs and LSTMs together. Our approach achieved first rank on all of the five English subtasks amongst 40 teams.