Sentiment for short texts using LSTM: Twitter
In this recipe, we will apply the LSTM algorithm to Twitter data, which we will classify by positive and negative sentiment. This will be similar to the Using LSTMs for supervised text classification recipe in the previous chapter. By the end of the recipe, you will be able to load and clean the data, and create and train an LSTM model for sentiment prediction.
Getting ready
For this recipe, we will use the same deep learning packages as before, and an additional package to segment Twitter hashtags, which can be downloaded at https://github.com/jchook/wordseg. After downloading, install it using this command:
python setup.py install
We also need to download the Twitter dataset, which can be found at https://www.kaggle.com/kazanova/sentiment140.
We will also use the tqdm package to see the progress of functions that take a long time to complete. Install it using the following:
pip install tqdm
