Order Matters at Fanatics Recommending Sequentially Ordered Products by LSTM Embedded with Word2Vec

TL;DR

This paper presents a novel e-commerce recommendation system that embeds products with Word2Vec and models their ordered sequence using a stateless LSTM RNN, improving click-through rates in production.

Contribution

It introduces the first recommendation system combining Word2Vec embeddings with LSTM for ordered sequences, applied in a large-scale real-world setting.

Findings

Outperforms previous Word2Vec-only recommender in click-through rate

First application of distributed Keras model predictions on Spark at this scale

Successfully models unidirectional product sequences in e-commerce

Abstract

A unique challenge for e-commerce recommendation is that customers are often interested in products that are more advanced than their already purchased products, but not reversed. The few existing recommender systems modeling unidirectional sequence output a limited number of categories or continuous variables. To model the ordered sequence, we design the first recommendation system that both embed purchased items with Word2Vec, and model the sequence with stateless LSTM RNN. The click-through rate of this recommender system in production outperforms its solely Word2Vec based predecessor. Developed in 2017, it was perhaps the first published real-world application that makes distributed predictions of a single machine trained Keras model on Spark slave nodes at a scale of more than 0.4 million columns per row.