摘要
深度學習近年來取得許多重要進展，本論文主要即在於將深度學習的技術應用於專利領域，特別是參考近期 NLP (自然語言處理) 及 NLG (自然語言生成) 的突破，本論文提出 “Augmented Inventing” (擴增發明) 的目標，希望透過機器協助發明人在專利領域提出更多發明。為了達到這個長遠的目標，本論文提出 “Augmented Patent Drafting Framework” (擴增輔助專利撰寫之實作框架) 做為短期可據以實施的基礎，並於研究期間完成數項相關實作。本論文中針對互動式專利文字生成提供的不同實作原型，可做為概念性驗證的研究參考。此外，本論文同時將深度學習的技術應用在專利領域過去習知的問題上，整體而言，在本論文實作過程中可以發現，深度學習領域的類神經網路模型不僅對習知的問題有效 (如專利分類)，對於如專利文字生成這類新的應用也呈現令人印象深刻的效果。本論文的主要貢獻包括數項達到最佳技術水平 (state-of-the-art) 的結果，其中部分項目在專利領域更為首創，因此，本論文有相當的可能性在專利領域中開創出新的子領域。最後，本論文的重要性日後將不限於專利領域，論文中相同的擴增輔助概念、深度學習技術、可據以實施的實作框架等，日後將有機會應用在不同的法學領域上。
並列摘要
Deep Learning has gained significant progress in recent years. This dissertation focuses on applying Deep Learning techniques to the patent domain. Particularly inspired by the contemporary NLP NLG breakthroughs, this dissertation proposes the "Augmented Inventing" goal, which is intended to help inventors be more inventive in the patent domain. For reaching the long-term goal, this dissertation proposes the "Augmented Patent Drafting Framework" as an obtainable reference implementation. This dissertation presents various prototypes of interactive patent text generation as a proof-of-concept. In addition, this dissertation addresses other conventional patent tasks by utilizing Deep Learning techniques too. Overall the implementations in this dissertation show that the neural network models in the Deep Learning field are effective for conventional patent tasks, such as patent classification, and impressive for new patent tasks, such as patent text generation. The contributions in this dissertation include several state-of-the-art results. Some of them are novel in the patent domain. From this perspective, this dissertation may have created a new sub-field for future researchers. Furthermore, the importance of this dissertation is not limited to the patent domain because the same augmentation idea, Deep Learning techniques, and reference implementation can apply to other legal domains in the future.
參考文獻
延伸閱讀
- 陳綺萱（2019）。深度學習之專利分析研究〔碩士論文，國立臺灣師範大學〕。華藝線上圖書館。https://doi.org/10.6345/THE.NTNU.GLIS.005.2019.A01
- 劉昌倫（2019）。深度神經網路模型在自動駕駛中的應用〔碩士論文，國立清華大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0016-0206202016144219
- 陳家駿（2019）。AI人工智慧機器／深度學習在藝術文化方面之應用暨其著作權適格保護與歸屬之探析。教育法學評論，(4)，1-28。https://doi.org/10.6920/ELR.201912_(4).0001
- 林世航（2018）。深度學習於進度管理之應用〔碩士論文，淡江大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0002-0209201815551500
- 蔣惟丞（2009）。The Development of a Rapid Machine Learning System〔碩士論文，國立虎尾科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0028-2407200919281700