社群網路中的意見形成賽局

Abstract

對於某個議題，每個人的看法與社群網路中的人際互動相關。本文旨在探討意見形成的過程，及其在均衡狀態下，由於共識未達成所產生的社會成本。第一部分，在特定條件的社群網路中，我們得到社會成本的上界。此上界與文獻中的結果相符合。第二部分提出演算法，透過指定幾個人表達特定意見，使社會成本降低，並衡量演算法優劣。

The Opinion Formation Games model the process how people form opinions in social networks, where the opinions are influenced by their friends. Based on the work of Bindel et al., every individual sees the lack of agreement as cost and aims to minimize it. The overall cost in networks could be extremely large. In this thesis, we first derive bounds on the price of anarchy for graphs with conditions and recover the previous results. Second, we give some algorithms to lower the social cost at equilibrium by controlling some individuals, and evaluate how good or bad they are.