高科技產業的技術變革迅速，若企業管理者能有效掌控前瞻性技術趨勢或競爭者技術脈動時，將可提前布局專利與提高競爭者技術進入障礙，保有領先的技術能力與市場。有效搶先布局專利技術已成為提升企業競爭力的關鍵要素，因此發展分析方法作為企業技術研發與專利布局的參考依據已成為重要議題。專利指標為簡易與有效的專利分析方法，透過多類型專利指標或變數形成視覺化的專利組合圖，受到企業管理者或研發人員廣泛使用。本研究發展二個創新分析模式的專利組合圖，以探討企業專利布局與研發策略，並以實證分析結果提供相關產業參考。模式一是利用資料包絡分析法的特性演算專利指標與變數並給予最適的權重分配，此模式可使企業自由選擇對自己評估最為有利的專利指標或變數的權重組合，並限制數據值介於0至1之間。模式一能有效提高專利指標或變數權重的客觀性，提升數據易判別性與精確性，並增加企業標竿學習的功能，將實證於磷酸鋰鐵電池專利技術並提供相關企業技術布局與研發規劃情報。模式二是整合對應分析與群集分析的視覺化方法，以呈現專利指標或變數間相對應的複雜關聯性，並藉由關聯性挖掘隱藏的專利情報。模式二能使專利指標或變數依據最原始的資料關聯性反映於二維圖形，可使企業簡易利用幾何空間距離遠近與密度的直觀視覺效果，有效探討與獲取專利情報，將實證於薄膜太陽能電池專利技術並提供相關企業技術布局與研發規劃情報。 The rate of technological progress in the hi-tech industry is increasing rapidly. If enterprise managers in industrial environments can identify the trends of emerging potential technology or the technological progress of its opponents, they could seize the opportunity to build patent fences and increase the entry obstacles of related technology. This, in turn, would allow them to maintain technological competitiveness and achieve dominance in technological capabilities or markets. Preempting the deployment of their patented technology could enhance the competitiveness of enterprises. Therefore, the development of analysis methods as reference for corporate research and development (R&D) and patent deployment has become an important issue. Patent indicators are simple and effective patent analysis methods that, through the use of multiple indicators or variables, form patent portfolio maps for enterprise managers or R&D personnel. This study develops two innovative models of patent portfolio analysis to explore the R&D planning and patent deployment of enterprise technologies and the practices in empirical patent analyses for related industries as a reference approach. Model 1 uses the characteristics of data envelopment analysis (DEA) to calculate the optimal weight set of patent indicators and variables for forming patent portfolio maps. This model allows enterprises to freely choose the patent indicator or variable weight set that is most advantageous by DEA evaluation, limiting the numerical values from 0 to 1. In addition, it increases the weight objectivity of indicators or variables, enhances the degree of distinguishability and accuracy, and increases the function of enterprise benchmark learning. Model 1 is evidenced in the lithium iron phosphate battery technology. Model 2 integrates the visualization methods of correspondence analysis and cluster analysis to visually present the effects of complex multiple patent indicators (or variables) corresponding with the relationships, which could be used in data mining to extract important patent information. This model allows enterprises to explore the associations among patent indicators or variables according to geometry distance or similarity, and to simultaneously visualize the results on a two-dimensional graphical display. Model 2 is evidenced in the thin film photovoltaic technology. These analysis results provide reference for patented deployment and R&D planning of enterprise technologies.
