This repository includes examples of how to:

1. Create item co-occurrence statistics from tabular data using Apache Beam
2. Learning item embeddings using TensorFlow tf.estimator APIs
3. Extract the trained embeddings and build an approximate nearest neighbours index using ANNOY