vanovermeire/keras-sentiment-analysis-examples

Example code accompanying my linkedin post on Dutch sentiment analysis.

Keras Machine Learning Examples

This project contains code accompanying my blog post on sentiment analysis with Keras.

The code is based both on official keras examples and the book Deep Learning With Python

Contents

This project contains:

• a dense model
• a covnet
• a GRU model

For sentiment analysis. It also contains code for creating a text-generation model.

The functions folder shows how you might encode files containing your text data and labels for use in Keras.

To save the tokenizer, pickle was used:

with open('example_tokenizer.pickle', 'wb') as handle:
    pickle.dump(tokenizer, handle, protocol=pickle.HIGHEST_PROTOCOL)

While for the model itself, I used the build-in save method:

model.save('example_saved_model.h5')

Notes on scraping

• The websites were scraped using a single-threader Python scraper with sleeps between the requests, to minimize any pressure on servers.
• Robots.txt files were respected.
• In deference to the websites, neither scraper nor raw data is included in this project.