Revealing Game Dynamics via Word Embeddings of Gameplay Data

Abstract

In this paper we show how word embeddings, a technique used most commonly for natural language processing, can be repurposed to analyse gameplay data. Using a large study of chess games and applying the popular Word2Vec algorithm, we show that the resulting vector representation can reveal both common knowledge and subtle details about the game, including relative piece values and the natural spatial flow of chess play. Our results suggest that word embeddings are a cheap and simple technique that can provide a broad overview of a game’s dynamics, helping designers and critics form new hypotheses about a game’s design, structure and flow.