近年來，隨著網路的發達，多人線上遊戲也越來越熱門。然而由於網路的隱匿性，遊戲中時常存在惡意玩家(使用者)為了私人的利益欺騙或擾亂其他玩家。玩家在遊戲中選擇一些其他玩家進行互動與合作，常因為對對方過去行為資訊的不足而做出錯誤的選擇與判斷，造成玩家的不安全感而大大減低線上遊戲的樂趣。因此，我們在這篇論文中提出一個適合各種線上遊戲的信任使用者選擇方法。在線上遊戲中，有許多影響玩家是否值得信任的因素，我們稱之為聲望因素(reputation factor)。聲望因素隨著遊戲的不同或使用者行為的不同而有所改變，其相對重要性也隨著改變。我們利用多變量分析(multivariate analysis)來計算每個聲望因素之權重，並且根據環境狀況做適當的權重調整，以便能夠讓玩家精確地選擇出最值得信任的一些其他玩家，以進行互動與合作。我們也針對所提的方法，執行許多模擬實驗來展示其效能。 In recent years, massively multiplayer online games (MMOGs) have become more and more popular. However, due to the anonymity of networks, malicious players (users) can cheat or disturb other players for their own benefits. Players may make wrong decisions about choosing players to interact or cooperate with because of the lack of accurate user selection schemes. In this thesis, we propose a trustworthy user selection scheme for MMOGs based on multiple user reputation factors. Different reputation factors have different relative importance. We utilize multivariate analysis to adjust the weights of reputation factors to reflect their importance. By the proposed scheme, an MMOG player can easily and accurately choose players that are most trustworthy to interact or cooperate with. We also perform simulation experiments for the proposed scheme to demonstrate its performance.
