HLRS: A Deep Reinforcement Learning-Based Hero Recommendation System for MOBA Games

Multiplayer Online Battle Arena (MOBA) games have gained immense popularity in recent years, with games like Dota2, League of Legends (LoL), and Honor of Kings (HoK) being the most popular. In a typical MOBA game, two teams consisting of 10 players battle against each other in two phases: hero selection and hero battle phase. Choosing a suitable virtual game character from all the available heroes during the hero selection phase is a daunting task due to the large number of current heroes, their complex relationships, and player interests. Therefore, we propose an innovative hero recommendation system called HLRS, which acts as a coach to guide players in selecting appropriate heroes during the selection phase. To verify the effectiveness of our model, we used HoK as a test platform and trained the model using its game data. We conducted numerous experiments, and the results showed that our model was more effective than other recommendation schemes. It analyzed various aspects to recommend the most suitable hero characters for players.