ENGLISH | JAPANESE

Final Results of the 3rd International AIWolf Competition at ANAC 2021

2021/08/26

Thank you for participating in the Third International AIWolf Competition. In this year, we had two divisions: Protocol Division and Natural Language Division.

Final Results of the Protocol Division

In the Protocol Division, 34 agents participated in the initial round, and 15 finalists were selected for the final round. The finalists played 400.000 werewolf games, and the results are as follows:

1st place: toku/ICE (both are the identical team)
2nd place: TOT
3rd place: KP22

Congratulations!

Ranking of all teams and winning rate per role
rank team revision win_rate villager seer medium bodyguard werewolf possessed lang
1 toku 197 0.635 0.797 0.820 0.827 0.806 0.293 0.301 JAVA
2 ICE 201 0.635 0.796 0.820 0.832 0.814 0.294 0.302 JAVA
3 TOT 223 0.625 0.800 0.776 0.803 0.836 0.300 0.256 C_SHARP
4 KP22 195 0.613 0.782 0.808 0.815 0.832 0.235 0.290 JAVA
5 SYu 57 0.609 0.785 0.794 0.824 0.850 0.216 0.291 JAVA
6 CanisLupus 232 0.604 0.772 0.815 0.818 0.836 0.208 0.288 JAVA
7 Tomatoken 203 0.602 0.776 0.762 0.824 0.788 0.229 0.284 JAVA
8 SORA 58 0.598 0.769 0.798 0.824 0.822 0.211 0.275 JAVA
9 Hideto 59 0.596 0.796 0.677 0.820 0.840 0.243 0.225 JAVA
10 HALU 231 0.595 0.785 0.702 0.833 0.814 0.248 0.232 PYTHON
11 Tomato 234 0.579 0.739 0.753 0.842 0.818 0.237 0.231 JAVA
12 OKAMI 123 0.577 0.793 0.632 0.764 0.755 0.208 0.225 PYTHON
13 karma 233 0.564 0.729 0.771 0.796 0.793 0.206 0.196 JAVA
14 wasabi 19 0.560 0.735 0.701 0.828 0.782 0.212 0.206 JAVA
15 Sashimi 236 0.538 0.764 0.566 0.733 0.794 0.140 0.207 JAVA

Final Results of the Natural Language Division

  • 1st Team: Udon (Japanese/English 1st)
  • 2nd Team: CanisLupus (Japanese/English 2nd)
  • 3rd Team: AzuEmon (Japanese/English 3rd)
  • Prof. Otsuki “Award: Full power idle talk” 全力雑談賞 Kanolab
  • Prof. Osawa “Award: Funny to play with” 遊んでいて楽しそうで賞 AzuEmon

Results of Subjective Evaluation

￭Average of human evaluations
￭Upper: Japanese Lower: English, via Google translate

  Lang Total A B C D E
AzuEmon JA 3.3 4.25 2.5 2.5 3.75 3.5
BIU Wolfs EN↔JA 2.35 3 2.25 2.25 2.25 2
CanisLupus JA 3.65 4 3.5 3.5 3.75 3.5
ForestLab JA 2.95 3.75 3 3.25 2.75 2
Kanolab JA 3 3.75 3 2.75 2.75 2.75
Udon JA 3.75 4 3.75 3.5 3.75 3.75
AzuEmon JA↔EN 3.38 3.67 3.56 2.89 3.00 3.78
BIU Wolfs-en EN 2.42 2.22 2.33 2.67 2.67 2.22
CanisLupus JA↔EN 3.42 3.67 3.33 3.33 3.67 3.11
Forestlab JA↔EN 3.07 3.33 2.78 3.56 3.33 2.33
Kanolab JA↔EN 3.20 3.67 3.50 3.17 3.00 2.67
Udon JA↔EN 3.87 4.00 4.67 4.25 3.00 3.87
Share

Leave a Reply

Your email address will not be published. Required fields are marked *