我國自施行專利法以來，就存在專利權事後廢棄（即舉發）的制度，且自始至今都不曾移除過。然而，卻鮮少有人分析我國專利權之舉發實務上的現況。除此之外，也鮮少有人有系統地探究如何有效地尋找系爭專利的舉發證據。對於專利權人而言，若能有效地挖掘舉發證據，可以儘早評估系爭專利遭舉發成立的風險，進一步避免日後投入大量的訴訟資源卻得到專利權無效的悲慘結局。對於非專利權人而言，則可以避免自家產品長期陷於專利權侵害的風險之中。因此，本研究之目的在於分析我國專利權之舉發案件的現況暨趨勢、以及各舉發案件之系爭專利與舉發證據之間的關聯性，藉以建立一套以專利文獻為舉發證據的檢索方法。
本研究採用「量化研究方法」分析我國專利權的舉發現況暨趨勢。根據中華民國專利資訊檢索系統的檢索結果，我國近5年智慧財產局一共作成755件舉發審定結果。本研究進一步將涉及專利權期間延長、司法機關審定、以及重複審定的舉發案件從分析對象中移除，故實際的分析對象為716件的舉發審定案件。本研究透過文書處理工具分析前述之716件的舉發審定案件，包括：發明專利權的被舉發次數、智慧財產局作成舉發審定的時程、舉發案件涉及的技術領域、舉發審定的結果、舉發對專利權的影響、以及舉發人提起舉發的事由等。另外，本研究進一步分析特定技術領域的舉發審定案件，包括：B25B、H01L、以及G06Q，並且發現當專利技術涉及G06Q時，其舉發成立的比例明顯較高。
有鑑於G06Q的舉發成立件數更勝於其他技術領域，故本研究以G06Q之舉發成立的案件作為關聯性分析的對象。透過逐一審閱各舉發案件的審定理由書，藉以找出對系爭專利具有影響力的舉發證據。本研究同樣採用「量化研究方法」分析系爭專利與舉發證據之間的關聯性，包括：引用脈絡、專利分類號、以及專利文本，藉以將具有關聯性的專利資訊轉換成檢索條件。基於關聯性與檢索效率的考量，本研究歸納出一套綜合式檢索方法，並且以B25B的舉發成立案件作為測試對象。最終，本研究發現該方法有高達94.12%的比例可以幫助使用者成功找到系爭專利的舉發證據，並且能幫助使用者將需要逐一審閱的專利文獻數量限縮至5.0%以內。

Since the enactment of Taiwan’s Patent Act, there has been a system of revocation of patent rights (a system of patent invalidation), and it has never been removed. However, few people have analyzed Taiwan’s current patent invalidation status, and studied the systematical methodology of searching invalidation evidences. If patentee can effectively search invalidation evidences, they can assess the risk of patent invalidation as soon as possible. Besides, they can avoid investing a lot of litigation resources and suffering the revocation of the patent rights in the future. For other people, they can avoid the risk of long-term patent infringement of its own products. In summary, the purpose of this study is to analyze Taiwan’s current patent invalidation status, explore the correlation between the patents at issue and their invalidation evidences, and establish a method of searching the patent documents as the invalidation evidences.
This study applies quantitative method to analyze Taiwan’s current patent invalidation status. According to the searching result of Taiwan Patent Search System (TPSS), Taiwan’s Intellectual Property Office (IPO) has issued 755 written decisions of patent invalidation in the past five years. Then the written decisions of patent invalidation involving extension of patent term, judicial decisions, and re-decisions are excluded in this study, so the precise number of subject to analyze is 716 written decisions of patent invalidation. These 716 written decisions of patent invalidation are analyzed by word processing tool in this study. The outcome of the analysis includes the number of patent invalidation per patent, the duration for the written decisions of patent invalidation by IPO, the technical fields of patent invalidation, the results of the written decisions of patent invalidation, the impact of patent invalidation on patent rights, and the ground for patent invalidation. In addition, this study further analyzed the written decisions of patent invalidation in specific technical fields such as B25B, H01L and G06Q. In conclusion, for the written decisions of patent invalidation in the filed G06Q, the proportion of successful results is obviously higher.
The number of successful result of patent invalidation in the filed G06Q is more than other technical fields, so this study uses these cases to analyze the correlation between the patents and the invalidation evidences. To analyze the correlation, this study distinguishes a substantial evidence from the invalidation evidences by reviewing every written decisions of patent invalidation. This study also applies quantitative method to explore the correlation between the patents at issue and the invalidation evidences, including citation context, patent classification numbers, and patent full texts. Based on the result of correlation analysis and search efficiency, a hybrid search method is developed in this study. In addition, the study applies the hybrid search method to test its effect on the successful result of patent invalidation in the filed B25B. In conclusion, the hybrid search method has a 94.12% success rate in searching the invalidation evidence, and can help users reduce the number of patent documents to be reviewed below 5.0%.

吳宜榛（2008），可專利性檢索之檢索技巧研究-以「專利工程師」為例，碩士論文，臺灣師範大學圖書資訊學研究所。
余杰謙（2008），利用文件摘要技術協助專利先前技術檢索，碩士論文，國立清華大學科技管理研究所。
陳科任（2010），結合申請專利範圍結構與專利說明的專利前案檢索方法，碩士論文，國立清華大學科技管理研究所。
陳嘉緯（2021），專利舉發證據實務-以I317967號專利為例，碩士論文，國立臺灣科技大學專利研究所。
曾元顯、高佐良、鄭浩（2011），中文專利前案檢索模式之成效評估，教育資料與圖書館學，第49卷第1期，頁75-102。
黃鳳梅（2007），應用專利申請範圍架構和發明元件關係以提升專利前案檢索之效能，碩士論文，國立清華大學科技管理研究所。
經濟部智慧財產局（2021），中華民國110年智慧財產局年報。
經濟部智慧財產局（2021），專利審查基準第五篇第一章。
蕭至淳（2020），以不變性圖像為基礎之專利搜尋與分類，碩士論文，國立臺灣大學資訊管理研究所。
Chin, A. (2008). Search for tomorrow: some side effects of patent office automation. NCL Rev., 87, 1617.
Shalaby, W., & Zadrozny, W. (2019). Patent retrieval: a literature review. Knowledge and Information Systems, 61(2), 631-660.
Tiwana, S. M. (2010). Methods for improving search on US patent documents. University of Southern California.
Tseng, Y. H., Lin, C. J., & Lin, Y. I. (2007). Text mining techniques for patent analysis. Information processing & management, 43(5), 1216-1247.