摘要
隨著知識經濟時代的來臨，智慧財產權愈來愈受到重視，專利品質的好壞因此倍受關注。過去多數研究指出，專利品質與專利被引證數關係密切，特別是專利在被引證時，具有資料截斷偏差問題，亦即一件剛核准專利於3至5年後，才會達到被引證高峰 [1] 。本研究以五個專利指標，技術生命週期、科學關聯性、技術成分、技術成分之相互依存度及專利引證數，經由統計學之複直線迴歸分析，來進行實務性的預測與探討。本研究結果顯示，於所選用之案例研究中，以複直線迴歸模型來預測專利被引證數是適合的；然而，於不同之案例研究中，影響迴歸預測專利被引證數之變數皆不盡相同，其意涵及應用都於本文中加以深入討論。
並列摘要
As the arrival of intellectual and economical age, the intellectual property rights are becoming more and more emphasized. This trend leads people to put more of their attention on the patent quality. In the past, many researches had suggested that the patent quality and the forward citation of a patent are closely related. Especially for truncation bias, the forward citations of a patent typically peak within at least three to five years from the issued date. This study adopts five patent indicators to evaluate the quality of a patent via multivariate linear regression. The preliminary results of various case patterns studied show that the model of multivariate linear regression is appropriate for the purpose. However, the variables used for forecasting the forward citation of a patent are not always the same among different case studies. Their implications and potential applications are discussed in depth in this research.