Functionality
Pre-Game Matchmaking
Matchmaking is the process of arranging queueing players into competitions.
(i.e. decide who are going to be teammates and who are going to be adversaries)
Fuxi Match provides the AI-enhanced matchmaking services to maximize the gross players’ satisfaction for upcoming matches.
AI algorithms are applied to 3 processes in a matchmaking system:
- User Modeling
- Arrival Modeling
- Online Planning
In-Game Win-Prediction & Comments-Generation
- Realtime win prediction with the match going on
- Realtime comments generation to summarize the match and highlight the key actions/moments
Post-Game Analysis
- Performance Rating
- Performance Radar Chart
- Piggybacking Detection
- Potential Cheating Diagnosis
- Power Leveling Detection
- Smurfing Detection
Matchmaking SaaS
Research & Blogs
Papers
- Fuxi AI Lab 【KDD’20】 OptMatch: Optimized Matchmaking via Modeling the High-Order Interactions on the Arena »Blog & Code & Data
- Fuxi AI Lab 【CIKM’20】Match Tracing: A Unified Framework for Real-time Win Prediction and Quantifiable Performance Evaluation »Blog & Code & Data
- Reproduction 【WWW’17】EOMM: An Engagement Optimized Matchmaking »Blog & Code & Data
Practical Tricks
- How to measure a win-prediction model in the matchmaking context
Basics and Prerequisites
- ELO
- TrueSkill