StarSpace: Embed All The Things!

TL;DR

StarSpace is a versatile neural embedding model capable of addressing various tasks like classification, ranking, recommendation, and graph embedding by learning similarities among entities with discrete features.

Contribution

It introduces a general-purpose neural embedding framework that is adaptable across multiple domains and tasks, outperforming or matching existing methods.

Findings

StarSpace achieves competitive results across diverse tasks.

The model is applicable to new problems beyond existing methods.

It effectively embeds entities with discrete features for similarity comparison.

Abstract

We present StarSpace, a general-purpose neural embedding model that can solve a wide variety of problems: labeling tasks such as text classification, ranking tasks such as information retrieval/web search, collaborative filtering-based or content-based recommendation, embedding of multi-relational graphs, and learning word, sentence or document level embeddings. In each case the model works by embedding those entities comprised of discrete features and comparing them against each other -- learning similarities dependent on the task. Empirical results on a number of tasks show that StarSpace is highly competitive with existing methods, whilst also being generally applicable to new cases where those methods are not.