專利的佈局技術是高科技產業在經濟市場中競爭之重要策略，而專利的申請及核準數往往更是一個國家競爭力的重要指標。在眾多的專利相關研究中，大多數的文獻為單獨研究某項專利指標或衡量各國家或企業的幾項專利指標後進行靜態分析，尚無作一整合性的討論。 生物科技(Biotechnology)為21世紀最具潛力之產業之一，相較於研發藥物的生技公司，生物晶片業進入門檻較低，本研究選定在生物科技與專利發展上具絕對領先地位的美國作為研究背景，欲探索美國生物晶片業各樣本企業在專利發展上的相對效能。在參考眾多文獻後選出：專利數目、專利成長率、專利被引證次數、引證指標、技術生命週期、科學關連性、技術集中度、衝擊指標、及技術影響力指標等九項專利指標進行深入分析、評選及指標修正，並藉此發展建立生技醫療產業之專利研發績效評估系統。 首先利用美國專利資料庫對各專利指標進行資料蒐集彙整與計算，再進行指標與研究樣本之因素分析、集群分析、與對應分析。其次利用文獻中的專利指標進行相對發展效能評估，成功地篩選出：專利數目、專利成長率、專利被引證次數、引證指標、及技術影響力指標等五項專利指標作為最終專利發展效能評估模型的指標，並得到科學關連性與技術集中度兩項指標對專利發展績效具有較顯著的影響。而後又發展出專利發展相對效能之標竿模型(benchmark model)作為評鑑新進企業的工具，建立四層效率前緣(efficient frontier)。 研究結果顯示在此產業中樣本企業在各項專利指標上具多元化的特色及差異，亦即專利發展策略定位迥異。透過不同方式的集群分析，本研究成功地將不同企業與不同專利指標進行分群。另外利用DEA資料包絡分析法及迴歸分析，本研究建立單一多面相專利發展績效指標並確認何種專利指標對專利績效發展效能值有顯著影響。最後，本研究參考標竿學習的學理並利用區別函數之觀念，建立專利發展效能的標竿模型並據此評估新進企業的專利研發績效區間及相對效能落點(location)，透過資料比對，確認此一標竿模型確實為可行，此模型在DMUs數極大時更能突顯出效能與實用性。 生物科技產業具有大量發展專利權以進行專利佈局(patent mining)進而提昇獲利的特色。本研究以專利分析及建立專利發展相對效能模型提供專利相關發展與定位的觀念，透過專利資訊的擷取，可從而制定出研發及產品策略。而此一分析模式未來也可運用在其它產業或進行產業間的比較上。 “Patent mining” is the essential strategy in the economy markets among high technology industries. Patent is also the important indicator for competitiveness of countries and companies showed in many researches. The relating issue of patent has been discussed extensively in academic and managerial oriented research for decades. But most researchers are focused on the unique patent indicator or the statistic analysis about the indicators of one nation or company. This paper will try to do the integrated research and establish a model to value the effectiveness of patent development. On the other hand, biotechnology is one of the most potential industries in 21st century. Biochip field has relatively low access barrier in biotechnology. It’s believed that Taiwan has relative competitive advantage to develop biochip based on the inherent electronics and micro-electromechanics groundwork. According to above background and motivation, this thesis chooses biochip d in the biotechnology as the research background. To explore the relative effectiveness in developing patents by the US biochip companies. We get nine patent indictors and modified some of them after reviewing references and gather the patent data from USPTO’s database. The descriptive statistics and qualitative analysis is given the article. We had a quantitative analysis thereafter; it’s the pioneer study in the patent researches. Secondly, this study uses DEA tool to establish the effectiveness valuing system. The system successfully separates the patent indicators we had in first part and chooses the real work variables. At last, we use benchmarking concept to form another evaluating model, the benchmark model is constructed by four efficient frontiers and gives an intuitive illustration concept. The results indicate there is large effectiveness indicators difference among 17 US biochip companies. It’s biased if the evaluating is based on only one or two indicators. Through factor and correspondence and cluster analysis, we group the companies and variables successfully. The study also provides a uni-dimension effectiveness valuing indictor through DEA method. And we develop a benchmark model to simplify and visualize the effectiveness evaluating process. This system works outstanding especially in large number of DMUs. This research is cascaded and linked between each steps. We provide both of quantitative and qualitative patent analysis. DEA and benchmark models are our two extraordinary contributions in this paper.
