高科技廠商專利策略與專利布局--以半導體封裝業為例

Abstract

根據美國專利局USPTO的統計，台灣近年來在美國申請的專利一直都維持在全球第四名，然而台灣從專利所獲得的收益卻遠不如其他排名在後的國家，原因在於其專利僅是在於「以量取勝」，而非以「以質超越」的專利布局。自80年代起便有許多經濟學者投入以專利指標作為專利質量衡量之研究。如經濟合作暨發展組織(OECD)、美國CHI Research公司所設計的指標衡量模式。但是專利指標僅能就一群專利賦予其統計數字，對個別廠商而言，無法明確得知本身的產品或技術是否侵犯他人專利。尤其台灣主要為代工產業，在技術上扮演跟隨者的角色，了解哪些專利對公司的產品有重大影響，進而採取相對應步驟，成為當前公司運作上實際而迫切的問題。本研究參考劉尚志博士提出的專利質量定義—專利被引證數與獨立項字數，利用文字探勘(Text mining)方式，針對一技術領域，找出針對目標公司產品在法律層面上權利範圍涵蓋最大的專利群，建構一個能反應真正影響該公司產品發展的專利布局，提供該公司作為未來商業策略的參考依據。 本研究的研究方法是以一家半導體封裝公司的某項產品為對象，探詢專家意見將覆晶封裝產品的技術分成數類(如：散熱、電性..等等)，再將此公司相關此產品的專利，按此分類分成不同群組，利用文字探勘工具針對不同群組進行分析，萃取出代表不同群組的字詞。再以這些字詞針對所設定的其他競爭者所擁有的專利進行比對分析，找出相似度高之專利後再根據專利質量強弱予以排序，進行專家審查，找出影響此公司覆晶產品發展的專利群。 利用這些專利所形成的專利布局機制，可以提供此一半導體封裝公司在覆晶技術上迴避或授權的參考，亦能提供公司未來商業策略的考量依據。

According to statistics of the United States Patent and Trademark Office (USPTO), the number of patents that Taiwan applied for in U.S.A. last year maintained the fourth place of the whole world once again. However, Taiwan gained profits from the patents far less than the countries got which ranked lower, resulted from “win with quantity” patents instead of “surmount with quality” patent portfolio. Since the 1980s a lot of economists have conducted researches on indexes for weighing the quality of patents, as those designed by Organization for Economic Co-operation and Development (OECD) and US CHI Research. But the patent indexes can merely give statistics on a group of patents. Manufacturers are unable to learn clearly whether their own products or technologies infringe others' patents. To Taiwanese manufacturers, which focus mainly on OEM or ODM manufacturing model and play the role of followers in technology, it has become practical and urgent in operation to understand which patents have great influence on their own products, and then adopt corresponding steps. This research reviews the definition of patent’s rating developed by Dr. Shang-Jyh Liu, and adopts the text mining approach. It aims at a certain technological field to locate the patent group which has great impact legally to a target company, and build up a patent portfolio that truly exerts influence on the company's products development, in order to offer the reference basis of commercial strategy in the future. This research method is to take a certain product of a semiconductor assembly company as the target. Based on specialists’ opinions, it first divides the technologies for products into several categories such as heatspreading, electricity, etc. Next text mining tools are utilized to analyze those categories and generate representative key words. And a comparative analysis on the target companies’ competitor’s patents is carried out to locate patents with close similarities and the patents are sorted based on the patents’ quality. Finally these patents are examined by expertees so as to determine which patents really impact legally the product development of the target company. Utilization of the patent portfolio design formed by these patents can provide this semiconductor assembly company not only with reference on technological avoidance and authorization, but also with a basis of future commercial strategy.