Werewolf game modeling using action probabilities based on play log analysis

Yuya Hirata, Michimasa Inaba, Kenichi Takahashi, Fujio Toriumi, Hirotaka Osawa, Daisuke Katagami, Kousuke Shinoda

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Citations (Scopus)

Abstract

In this study, we construct a non-human agent that can play the werewolf game (i.e., AI wolf) with aims of creating more advanced intelligence and acquire more advanced communication skills for AIbased systems. We therefore constructed a behavioral model using information regarding human players and the decisions made by such players; all such information was obtained from play logs of the werewolf game. To confirm our model, we conducted simulation experiments of the werewolf game using an agent based on our proposed behavioral model, as well as a random agent for comparison. Consequently, we obtained an 81.55% coincidence ratio of agent behavior versus human behavior.

Original languageEnglish
Title of host publicationComputers and Games - 9th International Conference, CG 2016, Revised Selected Papers
EditorsAske Plaat, Walter Kosters, Jaap van den Herik
PublisherSpringer Verlag
Pages103-114
Number of pages12
ISBN (Print)9783319509341
DOIs
Publication statusPublished - 2016
Externally publishedYes
Event9th International Conference on Computer and Games, CG 2016 - Leiden, Netherlands
Duration: 2016 Jun 292016 Jul 1

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10068 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th International Conference on Computer and Games, CG 2016
Country/TerritoryNetherlands
CityLeiden
Period16/6/2916/7/1

Keywords

  • Communication game
  • Multi-player
  • Player modeling
  • Werewolf game

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Werewolf game modeling using action probabilities based on play log analysis'. Together they form a unique fingerprint.

Cite this