SLIDE 1
Recommendation Applications and Systems at Electronic Arts
Meng Wu1 | John Kolen1 | Navid Aghdaie2 | Kazi A. Zaman2
1 EADP Intelligent Systems 2 EADP Data Platform
Recommendation Applications and Systems at Electronic Arts Meng Wu - - PowerPoint PPT Presentation
Recommendation Applications and Systems at Electronic Arts Meng Wu - - PowerPoint PPT Presentation
Recommendation Applications and Systems at Electronic Arts Meng Wu 1 | John Kolen 1 | Navid Aghdaie 2 | Kazi A. Zaman 2 1 EADP Intelligent Systems 2 EADP Data Platform IN INSP SPIRE IRE THE THE WORLD RLD TO PLA LAY EA DIGITAL PLATFORM
SLIDE 2
SLIDE 3
- Recommendation System
- Ap
Applica cations
- Challenges
- Conclusions
SLIDE 4
Recommendations everywhere
Out Outsi side
- Games f
for y r you
- Info f
from th the w web
Bet Between een
- Maps a
and m modes f for y r you
- Ma
Matchmak akin ing
Wi Within
- Dynamic d
difficulty ty
- Ob
Objecti tives
Recommendations As Next Best Activity
SLIDE 5
Outside
- Games for you
SLIDE 6
Outside
- News and info from the web
SLIDE 7
Between
- Maps and modes for you
8 game modes 20+ maps 4 character classes
SLIDE 8
Ne New players Team Death Match < Possession Ol Old and nd rec ecent ent players ers Team Death Match = Possession Old and inactive players Team Death Match > Possession
Game Mode Recommendation For Optimal Player Journey
Possession Team Death Match
SLIDE 9
Between
- Matchmaking
Last 3 3 Ou Outcomes Ch Churn Ra Rate DLW | | L LLW | | L LDW | |DDD 2.6% ~ ~ 2 2.7% … … WWW WWW 3. 3.7% … DLL | | L LWL | | L LDL 4.6% ~ ~ 4 4.7% WW WWL 4. 4.9% LLL LLL 5. 5.1%
Fair is not enough
SLIDE 10
Matchmaking for Maximal Entertainment
ℳ∗ = argmin
ℳ
* Pr(𝑞. 𝑑ℎ𝑣𝑠𝑜𝑡 | 𝒕., 𝒕9)
- <=,<> ∈ℳ
+ Pr(𝑞9 𝑑ℎ𝑣𝑠𝑜𝑡 | 𝒕9, 𝒕.)
SLIDE 11
Within
- Dynamic Difficulty Adjustment
LEVEL
0% 20% 10% 25 75 50 100 20 15 10 5 25
DIFFICULTY (#TRIALS) CHURN RATE
One EA match three game, 06/2016 – 07/2016
Churn Rate
- Avg. Trials to
Completion
SLIDE 12
Within
- Controlling Player Progression with DDA
2 3
Churn
4 5 1
Trial 1 Trial 2 Trial 3 Trial t
9 6 7 8
Easy Hard
Level 1 Level 2 Level 3 Level 4 Finished
Difficulty assignment for Level/Trial to maximize engagement
SLIDE 13
Generate engaging daily/weekly ”To Do” lists
- Make three goals
- Play Galactic War mode
- Apply a vehicle mod
Multiarmed bandit to maximize retention
Within
- Objectives
SLIDE 14
- Recommendation System
- Applications
- Ch
Challen enges es
- Conclusion
SLIDE 15
Challenges
- Games
- Players
- Data
SLIDE 16
Game Challenges Genres Franchises Extra Content
SLIDE 17
Player Challenges
Competitor Leader Storywriter
Motivations Acquisition
Or Organ anic Tr Trial Gu Guest Su Subscrib riber
SLIDE 18
Data Challenges
- Session days
- Duration
- Last player time
- Veteran
- Match count
- Skill
- Wins/loses (indiv)
- Goals
- Real Madrid fan
- Session days
- Duration
- Last player time
- Veteran
- Match count
- Skill
- Wins/loses (group)
- Kills/deaths
SLIDE 19
- Recommendation System
- Applications
- Challenges
- Conclusion
SLIDE 20
Our Strategy
- Systems over models
- Process over solutions
- One architecture to rule them all
SLIDE 21
Recommendation Architecture
Data Warehouse Recommendation Requests Game Servers Game Client Game Telemetry
SLIDE 22
Outline
- Recommendation System
- Applications
- Challenges
- Co
Conclusion
SLIDE 23
Conclusion
Conclusion
- Many opportunities for recommendation systems
in game industry beyond products.
- Recommendations as next best activity.
- Recommendation systems have an impact at EA
- Improved click-through-rate by 80% and player
engagement by 10%.
- EA studios share a unified recommendation
system platform
- Machine learning models as first class citizens
- Tightly coupled with experimentation
SLIDE 24