Player Engagement Prediction for Personalized Game Matchmaking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing matchmaking systems in multiplayer games fail to account for user-specific criteria that affect a user's engagement, leading to undesirable matches that reduce player enjoyment and engagement.
Innovation Solution
A method and system that utilize an engagement prediction model to analyze player data and match state data to predict player engagement, providing a predicted engagement metric as input to the matchmaking system to improve match decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional matchmaking systems pair players based on match state only, then matching speed and simplicity are maintained, but player engagement and enjoyment deteriorate
Solution Approach 1:
The system performs preliminary analysis of player data and match state data before making matchmaking decisions. An engagement prediction model is trained in advance on historical data to predict player engagement metrics, allowing the system to evaluate multiple potential matches efficiently without complex real-time calculations during the actual matchmaking process.
Solution Approach 2:
The patent introduces an engagement prediction model as an intermediary component between the traditional matchmaking system and player data. This model acts as a mediator that processes player characteristics and match state information to generate engagement predictions, which then guide the matchmaking decisions, simplifying the overall system architecture while improving engagement.
2Productivity
If players are matched to undesirable game scenarios, then match state balance is maintained, but player retention and continued engagement worsen
Solution Approach 1:
The system implements feedback mechanisms by continuously monitoring player engagement metrics and using this information to refine future matchmaking decisions. The engagement prediction model is trained on historical player behavior data, creating a feedback loop where past player responses to different match scenarios inform future matching decisions, progressively improving match quality and player retention.
Solution Approach 2:
The patent changes the parameters used for matchmaking from purely match-state-based criteria to include predicted engagement metrics derived from player-specific characteristics. By transforming the matchmaking parameters to incorporate engagement predictions, the system optimizes for both match quality and player retention without compromising match state balance.
3Loss of time
If players join matches after the game is already in progress, then match formation speed is improved, but player engagement and enjoyment deteriorate
Solution Approach 1:
The system performs preliminary evaluation of how late-joining players will impact match engagement using the trained prediction model. By assessing engagement risks before finalizing the match, the system can make informed decisions about whether to allow late joiners, thereby minimizing negative impacts on overall player engagement while maintaining reasonable match formation speeds.
Data Source
AI summary
A method for matchmaking of game players includes receiving player data associated with a user account, receiving match state data associated with a match of the game, and based on the player data and the match state data, extracting engagement prediction features. The method further includes providing, as an input to an engagement prediction model, the engagement prediction features, and receiving, as an output from the engagement prediction model, a predicted engagement metric, the predicted engagement metric being based on the engagement prediction features. The method further includes providing the predicted engagement metric as an input to a matchmaking system and receiving from the matchmaking system a decision whether to match the user account to the match of the game for gameplay.