- 學位論文
引證關係於專利檢索之應用研究
Applying Patent Citation Network to Reinforcement of Patent Search Efficiency
摘要
有鑑於利用關鍵字搜尋於專利檢索時往往需要人工花大量時間來做判讀，費時費力，且一般在使用關鍵字進行專利檢索時較難同時兼顧回收率和準確率等原因，本研究提出一種利用專利引證關係網路來補強專利檢索結果之方法，藉以使專利檢索的廣度增加而不降低其原有的準確率。此方法主要可分為三個階段，第一階段係將美國專利資料庫(USPTO)中之特定檢索標的利用專利引證關係網路作分析，並以共引證分析的概念過濾出一群與檢索標的相關聯之候選專利群；在第二階段中，本研究方法以國際專利分類號(IPC)做為技術領域的判斷依據，加以計算出候選專利群中之各專利與各檢索標的專利間之相似度，進而建立一相似度矩陣；第三階段係針對該相似度矩陣進行資料處理與分析，在候選專利群的各專利中篩選出與檢索標的最為相似之專利。最後，本論文提供兩個實際的案例研究做為驗證，亦證實利用本論文之方法可提高專利檢索之效率。
並列摘要
The quality of the patent searches is key importance to the patent analysis. However, the conventional method for patent searches is time-consuming and can hardly give consideration to both recall ratio and precision ratio. This thesis proposes a method of reinforcing the result of patent searches and improving the efficiency, thus improving the recall ratio without affecting the precision ratio thereof. More specifically, the method comprises three phases: Phase I - using the USPTO database to generate a group of candidate patents by applying patent citation network and co-citation analysis; Phase II - establishing a similarity matrix by comparison between the original patents and candidate patents; and Phase III - data analysis and sorting of the similarity matrix, so as to find out patents which are most similar to the original patents from the candidate patents. Furthermore, the thesis takes Through-Silicon Vias (TSVs) and Silicon based Biochip as two study cases to verify the feasibility and effectiveness of the proposed method in this thesis.