Coalition Formation among Agents in Complex Problems Based on a Combinatorial Auction Perspective

摘要

In this paper, we try to apply a combinatorial auction protocol to a coalition formation among agents to solve complex problems. This paper focuses on a scheduling problem that considers various constraints as a complex problem. Constraints on scheduling can be expressed as combinations of items (time slots) in a combinatorial auction. Agents bid for necessary combinations of time slots to satisfy users' requirements. We formalize a combinatorial auction for scheduling as an MIP (Mixed Integer Programming) problem, which integrates the constraints on items and bids to express complex problems. We have experimentally confirmed that our method can obtain a socially preferable schedule in practical time.