辨識專利侵權訴訟關係網絡中位置與角色的計量分析方法：以智慧型手機公司為例

Abstract

專利侵權訴訟的競合關係，如何決策？專利侵權訴訟網絡中位置與角色如何辨識？為專利侵權訴訟重要的管理課題。本文以智慧型手機專利侵權訴訟關係的網絡為例，提出系統 性之分析程序。首先，依陡坡法則篩選 56 間訴訟公司為分析樣本，將訴訟關係建構具方向性 訴訟關係網絡，透過主成分分析將向外程度中心性、特徵向量中心性和向內程度中心性，萃 取成主動性與被動性二個集群變數。並且採用兩階段集群分析：一、利用 MANOVA 檢定與 變異數比率準則，決定最合適之分群數目為 4；二、使用已知 4 群組數，以主動與被動為集群 變數進行 K-means 集群分析，並以 MANOVA 檢定 4 個位置的顯著性差異。另建構位置內與位置間之訴訟關係強度與強度顯著性檢定之方法與指標，建構位置內與位置間之訴訟路徑，並依每一位置的訴訟關係與位置內公司的共同性質，對每一位置角色的行為訂定合宜的標 名。

How to make strategic decision on competitive relation of patent infringement litigation? How to recognize the position and role on patent infringement litigation network? These are important issues of management. This article brings out a systematic analytic process on the patent infringement litigation network of smart phone as an example. First, the sample contains 56 companies after selecting based on Screen Test. Companies’ litigation relations construct a directed litigation relation network. In-degree centrality, out-degree centrality and eigenvector centrality of network are converted into two variables of activeness and passiveness by Principal Component Analysis. Two-stage Cluster analysis contains the first stage that decides the cluster counts is 4 by 1-way MANOVA and Variance Ratio Criteria and the second stage that classifies companies into 4 clusters by activeness and passiveness as cluster variables of K-means analysis and tests significance of these 4 positions by MANOVA. In addition, this research builds an indicator for significant test of litigation frequency by Frequency within Position and Frequency between Positions. Finally, the role of each position is labelled by the paths among these 4 positions and the common characters of companies within each position.