Author: Aleo Aninda
Aninda, Aleo, 2019 Meta discovery and role-based matchmaking (MEDIROMA), Flinders University, College of Science and Engineering
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
Player engagement and satisfaction has been one of the top priorities for game developers of competitive multiplayer games. One of the challenges for the developers is to ensure the balance between fairness and satisfaction for every match that the player may experience. However, ensuring satisfaction can be difficult since many factors play into how the player experiences the game. For example, a player may encounter team mates who would get into a conflict for having an overlapping choice of roles within the game. This would result in either a compromise or disagreement that would lead to decreasing the team’s chances of winning. However, this can be avoided if the matchmaking solution takes into account of the player’s role preference before placing them into the teams. This would avoid any arguments and give the players more satisfaction with their games. To achieve such a solution, the developer must identify the roles that fit into the structure of the game that guarantees a moderate to high chance of winning a game. Such a role composition is defined by the Most Effective Tactical Advantage (META) of the game, which can be mined from existing dataset of match statistics for that game. The research presented here proposes a model which utilises the META of the game and allocates players based on their role preference. This solution has been achieved by using feature space exploration routine, such as Mean Shift clustering algorithm and identifying the cluster that represents the META of the game. The META information from that cluster is then passed to the matchmaking solution, which then leverages that information to allocate players by looking at their past matches to predict their preferred role. This overall model would ensure player satisfaction along with fairness of the game. The model was applied on League of Legends, using a match statistics dataset of over six thousand matches. It produced a META cluster, which has a composition of roles that has an 87.65% chance of winning a game. Afterwards, a positive feedback was received from League of Legends players who attempted to follow the role composition given by the META cluster. Based on this feedback, it can be postulated that the model was able to deliver on its promised goals of improving player satisfaction, and this can be emulated in the live build of a competitive game like League of Legends.
Keywords: matchmaking, video games, meta, league of legends, role based
Subject: Computer Science thesis
Thesis type: Masters
Completed: 2019
School: College of Science and Engineering
Supervisor: Dr Brett Wilkinson