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

doi:10.1007/978-3-319-50935-8_10

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 proceeding › Conference 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 language	English
Title of host publication	Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers
Editors	Aske Plaat, Walter Kosters, Jaap van den Herik
Publisher	Springer Verlag
Pages	103-114
Number of pages	12
ISBN (Print)	9783319509341
DOIs	https://doi.org/10.1007/978-3-319-50935-8_10
Publication status	Published - 2016
Externally published	Yes
Event	9th International Conference on Computer and Games, CG 2016 - Leiden, Netherlands Duration: 2016 Jun 29 → 2016 Jul 1

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10068 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	9th International Conference on Computer and Games, CG 2016
Country/Territory	Netherlands
City	Leiden
Period	16/6/29 → 16/7/1

Keywords

Communication game
Multi-player
Player modeling
Werewolf game

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-319-50935-8_10

Cite this

Hirata, Y., Inaba, M., Takahashi, K., Toriumi, F., Osawa, H., Katagami, D., & Shinoda, K. (2016). Werewolf game modeling using action probabilities based on play log analysis. In A. Plaat, W. Kosters, & J. van den Herik (Eds.), Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers (pp. 103-114). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10068 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-50935-8_10

Werewolf game modeling using action probabilities based on play log analysis. / Hirata, Yuya; Inaba, Michimasa; Takahashi, Kenichi et al.

Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers. ed. / Aske Plaat; Walter Kosters; Jaap van den Herik. Springer Verlag, 2016. p. 103-114 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10068 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Hirata, Y, Inaba, M, Takahashi, K, Toriumi, F, Osawa, H, Katagami, D & Shinoda, K 2016, Werewolf game modeling using action probabilities based on play log analysis. in A Plaat, W Kosters & J van den Herik (eds), Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10068 LNCS, Springer Verlag, pp. 103-114, 9th International Conference on Computer and Games, CG 2016, Leiden, Netherlands, 16/6/29. https://doi.org/10.1007/978-3-319-50935-8_10

Hirata Y, Inaba M, Takahashi K, Toriumi F, Osawa H, Katagami D et al. Werewolf game modeling using action probabilities based on play log analysis. In Plaat A, Kosters W, van den Herik J, editors, Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers. Springer Verlag. 2016. p. 103-114. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-50935-8_10

Hirata, Yuya ; Inaba, Michimasa ; Takahashi, Kenichi et al. / Werewolf game modeling using action probabilities based on play log analysis. Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers. editor / Aske Plaat ; Walter Kosters ; Jaap van den Herik. Springer Verlag, 2016. pp. 103-114 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{b28707c6c3db4e50ae62ee8f1f3d32b5,

title = "Werewolf game modeling using action probabilities based on play log analysis",

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.",

keywords = "Communication game, Multi-player, Player modeling, Werewolf game",

author = "Yuya Hirata and Michimasa Inaba and Kenichi Takahashi and Fujio Toriumi and Hirotaka Osawa and Daisuke Katagami and Kousuke Shinoda",

note = "Funding Information: We heartily thank Mr. Ninjin for allowing us to use the data of the Werewolf BBS. This study received a grant of JSPS Grants-in-aid for Scientific Research 15K12180. Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 9th International Conference on Computer and Games, CG 2016 ; Conference date: 29-06-2016 Through 01-07-2016",

year = "2016",

doi = "10.1007/978-3-319-50935-8_10",

language = "English",

isbn = "9783319509341",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "103--114",

editor = "Aske Plaat and Walter Kosters and {van den Herik}, Jaap",

booktitle = "Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers",

}

TY - GEN

T1 - Werewolf game modeling using action probabilities based on play log analysis

AU - Hirata, Yuya

AU - Inaba, Michimasa

AU - Takahashi, Kenichi

AU - Toriumi, Fujio

AU - Osawa, Hirotaka

AU - Katagami, Daisuke

AU - Shinoda, Kousuke

N1 - Funding Information: We heartily thank Mr. Ninjin for allowing us to use the data of the Werewolf BBS. This study received a grant of JSPS Grants-in-aid for Scientific Research 15K12180. Publisher Copyright: © Springer International Publishing AG 2016.

PY - 2016

Y1 - 2016

N2 - 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.

AB - 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.

KW - Communication game

KW - Multi-player

KW - Player modeling

KW - Werewolf game

UR - http://www.scopus.com/inward/record.url?scp=85007349294&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85007349294&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-50935-8_10

DO - 10.1007/978-3-319-50935-8_10

M3 - Conference contribution

AN - SCOPUS:85007349294

SN - 9783319509341

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 103

EP - 114

BT - Computers and Games - 9th International Conference, CG 2016, Revised Selected Papers

A2 - Plaat, Aske

A2 - Kosters, Walter

A2 - van den Herik, Jaap

PB - Springer Verlag

T2 - 9th International Conference on Computer and Games, CG 2016

Y2 - 29 June 2016 through 1 July 2016

ER -

Werewolf game modeling using action probabilities based on play log analysis

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this