以主路徑分析與多準則決策分析法探勘第三代半導體專利

Abstract

寬能隙 (Wide-Bandgap, WBG) 半導體也稱為第三代半導體 (Third-Generation Semiconductor)，近年來於工業4.0、電動車、5G行動通訊與可再生能源等高科技領域發揮重要且不可替代的作用。目前，第三代半導體技術仍然快速發展，但主流技術尚不成熟。因此，了解過去與目前技術發展及佈局狀況非常重要，可作為專利探勘 (Patent Mining) 的結果，可為未來廠商與學術、研究單位開發新產品提供寶貴的資訊，但少有相關研究發表。為解決本問題，本論文擬提出一種新型分析架構，整合主路徑分析、多準則決策分析(Multi-Criteria Decision Making，MCDM)與多目標決策分析法 (Multi-Objective Decision Making，MODM) ，分析由美國專利商標局 (United States Patent and Trademark Office，USPTO) 專利資料庫探勘之第三代半導體專利。首先，本研究使用美國專利商標局 (USPTO) 資料庫探勘所得之61,029項第三代半導體專利，使用最長路徑分析 (Longest Path Analysis)，獲得由24項專利組成的最長主路徑，並透過搜尋路徑數 (Search Path Count，SPC) 主路徑分析，得出由26項專利建構而成的四條主路徑，並將所有專利引用結果導入決策實驗室分析法 (Decision Making Trial and Evaluation Laboratory，DEMATEL) 之初始影響矩陣，並利用基於決策實驗室之網路流程法 (DEMATEL based Analytic Network Process，DANP)， 計算出各專利之間的相互影響權重，並推導影響程度最高之專利與其影響關係。本研究選取以最長路徑所得之源點 (source node)，影響程度最高之專利、搜尋路徑數法所得，四條主路徑上的26項專利技術、與使用影響關係圖 (Influential Relationship Map，IRM) 所得，影響權重最高之專利所對應之技術，為個案公司之研發能力集合 (Competence Set)，並由模糊能力集合擴展法，求得最小生成樹 (Minimum Spanning Tree)。最後，經確認適合發展時程後，可發展為第三代半導體的技術路徑圖。本研究所發展之分析架構，可作為未來各研究機構與科技公司探勘專利、定義技術路徑圖之用，所得第三代半導體專利探勘之結果，亦可作為未來發展第三代半導體技術之依據。Wide-Bandgap (WBG) Semiconductor, also known as Third-Generation Semiconductor, has played a crucial and irreplaceable role in recent years in high-tech fields, such as Industry 4.0, electric vehicles, 5G mobile communications, and renewable energy. Currently, the development of Third-Generation Semiconductor technology is still rapidly advancing, but mainstream technologies are not yet mature. Thus, understanding the past and current technological developments and trends is of utmost importance and can serve as valuable insights for patent mining. The results of this exploration can provide valuable information for future manufacturers, academic and research institutions in developing new products, though there is limited existing research on this subject. To address this issue, this paper proposes a novel analytical framework that integrates Main Path Analysis (MPA), Multi-Criteria Decision Making (MCDM), and Multi-Objective Decision Making (MODM) techniques to analyze Third-Generation Semiconductor patents obtained from the United States Patent and Trademark Office (USPTO) patent database.Firstly, this study retrieved 61,029 Third-Generation Semiconductor patents from the patent database of USPTO. Then, a longest path composed of 24 patents, four main paths with 26 patents and a influence relationship map (IRM) consisting of 62 patents were derived by using the longest path analysis, the search path count (SPC) based MPA, as well as the Decision Making Trial and Evaluation Laboratory (DEMATEL). Using the DEMATEL based Analytic Network Process (DANP), inter-patent influence weights are calculated, and the most important patents and its causal relationships are determined.This study selects the source node derived from the longest path, the most influential patent derived by using the DEMATEL, the 26 patents from the four main paths, to form a competence set for fast catching-up firms of Third- Generation Semiconductors. Through the expansion of the fuzzy competence set, the Minimum Spanning Tree (MST) is obtained. Finally, after confirming an appropriate developmental schedule, the result can serve as the technological roadmap for technology firms of Third-Generation Semiconductor. The analytic framework developed in this thesis can serve as a basis for research institutions and technology firms for technology mining and roadmap definitions. The results of technology mining can also serve as a foundation for the future development of Third-Generation Semiconductor technology.