近年來，群體排名的問題已經變得越來越重要。然而，在大部份群體排名的應用中，皆著重於如何從所有使用者的排名資料，找到最多人所擁有共識的排序；而較少有研究是著重在找出那些持有不同於主流看法的排序。因此，本論文欲針對此問題進行研究。在本篇論文中，若針對同一個排序，其贊成此排序的人數若與反對此排序的人數相當，我們即將此排序定義為爭議性排序。藉由找到這些爭議性排序，決策者可以瞭解為何會有不同於主流意見的衝突或矛盾的情形發生，甚至可以更進一步瞭解其有哪些支持者與反對者。藉由我們所挖掘到的爭議性排序，決策者可以嘗試與這些持有不同意見的人做溝通，或是試著解決他們的疑惑，讓整個公司可以更和諧，運作得更為順利。 在本篇論文中，我們會提出一個演算法幫助我們從使用者的意見中，找到具有爭議性的排序。我們也會使用人工資料集和真實資料做實際的實驗測試。而我們實驗的結果指出，我們所提出的方法是一個效率高，並且能夠定義爭議性意見、並應用在實際的決策。 ;In recent years, the group ranking problem has become an important study. In most of group ranking problems, the focuses lie on finding the consensuses upon which most people agree. No previous researches have paid attention on finding conflict opinions, called conflict patterns in this work, among decision makers. In this work, we define conflict patterns as those orderings of alternatives which have roughly the same numbers of pros and cons. The conflict patterns can reveal the ranking of what alternatives are the most controversial among decision makers, and who are supporters and opponents. With the information of conflict patterns, we can communicate with those people with different opinions and try to resolve the differences. In our work, an algorithm, MCP, is developed to find these conflict patterns from users’ partial ranking data. Extensive experiments are carried out using synthetic data sets and real data. The results indicate that the proposed method is computationally efficient, and can effectively identify conflict patterns among all users.
