Using Collaborative Filtering to Recommend Champions in League of Legends

TL;DR

This work addresses champion recommendation in League of Legends by formulating a collaborative-filtering approach using $SVD$ to map players and champions into a shared latent-factor space. Champion Mastery Points (normalized to 1–100) serve as the proxy for user preferences, enabling training on a dataset of 2514 Riot API players. A preliminary user study with 30 participants demonstrates that system-provided recommendations are rated significantly higher than random ones, supporting the viability of the approach for enhancing player engagement. The paper also discusses limitations related to metagames and champion updates, compares alternative algorithms, and outlines future work including parameter tuning and more extensive evaluation.

Abstract

League of Legends (LoL), one of the most widely played computer games in the world, has over 140 playable characters known as champions that have highly varying play styles. However, there is not much work on providing champion recommendations to a player in LoL. In this paper, we propose that a recommendation system based on a collaborative filtering approach using singular value decomposition provides champion recommendations that players enjoy. We discuss the implementation behind our recommendation system and also evaluate the practicality of our system using a preliminary user study. Our results indicate that players significantly preferred recommendations from our system over random recommendations.

Figures (2)

• Figure 1: Sample survey given to a user where * indicates a random recommendation. This is not visible to the subject
• Figure 2: Histogram of user rated scores given to random recommendations