Study on Problem-solving by Organizing Humans and Machines

2022 Fiscal Year Final Research Report

• Project/Area Number: 19H04170
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Review Section: Basic Section 61030:Intelligent informatics-related
• Research Institution: Osaka University (2020-2022); Kyoto University (2019)
• Principal Investigator: Matsubara Shigeo 大阪大学, 数理・データ科学教育研究センター, 特任教授(常勤) (80396118)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: マルチエージェントシステム / 集合知 / インセンティブ設計

Outline of Final Research Achievements

Toward establishing a problem-solving method through the organizational formation of people and machines, we have developed the idea of the diversity prediction theorem, a basic theory of collective intelligence, from prediction problems to various problems. From the viewpoint that the performance exhibited by a group is determined by the abilities and diversity of individuals, we examined the promotion of cooperation and the suppression of collusion. Specifically: (1) As a prediction diversity maintenance method, we have devised an efficient task partitioning method for sequential crowdsourcing. (2) As a prediction diversity expansion method, we have devised a decentralized regulation method based on contracts among participants to resolve the imbalance between ride demand and vehicle allocation in free-float car sharing. (3) Toward devising an individual error reduction method, we have conducted an equilibrium analysis of bribery in a reputation system on EC sites.

Free Research Field

知能情報学

Academic Significance and Societal Importance of the Research Achievements

本研究は、人と機械の協働を人と機械の組織形成による問題解決と捉え、クラウドソーシングにおけるタスク割当て、カーシェアリングの分散制御、評判システムの戦略的操作など予測問題と異なる問題領域でも多様性予測定理の考え方に基づく問題解決が有効であることを示した。これは、これまで別々に扱われてきた問題を人と機械の組織形成という観点から統一的に取り扱える可能性を示すもので、集合知の理解を一歩進めるものとなっている。本成果は、人工知能に関する最難関会議での採録や、エージェントに関する国際会議での最優秀論文賞の受賞に見られるように高い評価を受けている。