本研究首先以水解基板製程加工，建立以結合創新之餘弦相似度方法之創新專利搜尋法及專利分析探討，並進一步結合創新之餘弦相似度概念改善專利歸屬(PJ值)之機率值方法。本文之創新餘弦相似度方法，其主要為將文件中所有的重要詞彙都視為一個個的向量維度，以該詞彙的權重為該維度的值，組合而成一向量。本研究改良式餘弦相似度提出以各技術字、零組件元件字及功能字之常態化數值當成該詞彙之權重。將各技術及功能高度相關之專利歸類該群組字群，接著投入一新專利並得到該專利技術及功能字群，利用改良式餘弦相似度方法進行與高度相關專利群組字群比對，比對完成後進而擴充該技術及功能層之高度相關字群。擴充建立高度相關字群後，在進行專利歸屬(PJ值)之機率值方法前可以先利用改良式餘弦相似度比對的方式先剔除掉不相關的技術層或功能層，達到較準確且較快速的判斷專利歸屬方法。上述方法針對水解基板製程加工建立結合餘弦相似度改良專利搜尋方法及專利歸屬之機率值並進行測試完成後，再以LED閱讀燈為研究之載具，結合改良式餘弦相似度之專利搜尋法檢索出LED閱讀燈之繁體和簡體中文及英文專利文件，並進行LED閱讀燈繁體和簡體中文及英文專利的IPC分析。再利用改良式餘弦相似度比對繁體和簡體中文及英文專利之技術字和零組件元件字及功能字之同義字進行中英文專利搜尋。本文由繁體和簡體中文及英文專利語意分析及斷詞斷字系統結合改良式餘弦相似度概念，半自動半人工歸納出相關繁體和簡體中文及英文專利之技術及零組件字群與功能字群並統計各技術及零組件字與功能字出現次數計算各常態化數值。接著應用所搜集的LED閱讀燈相關繁體和簡體中文及英文專利，收集同義字，建立繁體和簡體中文及英文同義字對照表。接著建立LED閱讀燈相關繁體和簡體中文及英文專利之第一層及第二層技術之技術/功能矩陣，並進行以技術觀點為主的LED閱讀燈繁體和簡體中文及英文專利分析。本文先應用判斷以水解基板製程加工專利所歸屬的技術或功能類別之方法，再投入LED閱讀燈相關專利。利用上述建立的繁體和簡體中文及英文同義字表，配合技術零組件元件字、技術字群及功能字群之關鍵字之出現次數，建立關鍵字相關的模糊值，進而應用次數常態化概念，先以改良式餘弦相似度做篩選再以(PJ值)機率方法計算出該繁體和簡體中文及英文專利可能歸屬之技術或功能類別的概率值。
接著本文以完成整理之LED閱讀燈的繁體和簡體中文及英文專利進行IPC分析、發明人分析、公司別分析、所屬國別分析以及整體趨勢分析。並由繁體和簡體中文及英文專利之元件字、技術詞和功能詞的相關斷詞及斷字分析，建立技術功能類別之字群，進而產生第一層及第二層技術功能矩陣。本文進而由第一層及第二層技術之技術/功能矩陣建立第一層及第二層各分項技術與公司別、核心競爭力公司分析圖、競爭者研發活動力趨勢圖及技術層活動力趨勢圖。最後實際投入相關專利驗證其歸屬之第一層及第二層技術類別，證明本研究之方法為可行。上述所搜尋的相關LED閱讀燈的繁體和簡體中文及英文專利，以及前述之IPC分析和技術分析之結果，可提供給企業及工程師作為參考，以節省研發時間，作為新專利產生之參考，此外亦可使企業了解競爭者之核心競爭力。

The study firstly aims substrate hydrolysis process to establish an innovative patent search method and patent analysis that combine with the innovative cosine similarity method. Furthermore, the study combines with the innovative concept of cosine similarity to improve the probability of patent attribution (PJ value). For the paper’s innovative cosine similarity method, all the important terms of a patent document are regarded as individual vector dimensions. The values of different dimensions, being the weights of those terms, are combined as a vector. Improved cosine similarity developed by the study proposes treating the normalized numerical values of different technical words, part and component words and functional words as the weights of those terms, and categorizing the patents highdy relating to various techniques and functional in a word cluster.After that, the study adds a new patent, and acquires the cluster of technical and functional words of this patent. Using improved cosine similarity method, the study compares the word clusters of highdy-related patents. After comparison, the highdy-related word clusters of technical and functional hierarchies are expanded. After expansion and establishment of highdy-related word clusters, and before using the probability method of patent attribution (PJ value), the comparison way of improved cosine similarities can be firstly employed to rule out the irrelevant technical hierarchy or functional hierarchy so as to achieve a more accurate and faster judgment method of patent attribution. After completion of establishment and test by the above methods for substrate hydrolysis process, the study uses LED reading lamp as a carrier of the research, combines with the patent search method of improved cosine similarity to retrieve the traditional and simplified Chinese and English patent documents in relation to LED reading lamp, and conducts IPC analysis of the traditional and simplified Chinese and English patents of LED reading lamp. Then the study uses improved cosine similarity to compare the technical words, part and component words, and synonyms of functional words in the traditional and simplified Chinese and English patents, and search the relevant Chinese and English patents. Combining the semantic analysis of the traditional and simplified Chinese and English patents and the term segmentation and word segmentation system with the concept of improved cosine similarity, the paper semi-automatically and semi-manually induces the technical and component word cluster and functional word cluster of relevant traditional and simplified Chinese and English patents, makes a statistics of the frequency of various technical words, component words and functional words, and calculates their normalized numerical values. Subsequently, the study applies the searched traditional and simplified Chinese and English patents in relation to LED reading lamp and collects the relevant synonyms in order to establish a check list of the traditional and simplified Chinese and English synonyms. After that, the study establishes the first and second hierarchies of technical/functional matrices of the traditional and simplified Chinese and English patents relating to LED reading lamp, and analyzes the technical viewpoint-oriented traditional and simplified Chinese and English patents of LED reading lamp. First of all, the paper applies the method of judging whether a patent of substrate hydrolysis process belongs to technical or functional category, to patents in relation to LED reading lamp. Using the check list of traditional and simplified Chinese and English synonyms established above, together with frequency of the keywords in part and component word cluster, technical word cluster and functional word cluster, the paper establishes fuzzy values for relevant keywords, and further applies the concept of normalized frequency. The paper firstly uses improved cosine similarity to make screening, and then adopts probability method (PJ value) to calculate the probability for the traditional and simplified Chinese and English patents to belong to technical or functional category. Then, focusing on the rearranged traditional and simplified Chinese and English patents of LED reading lamp, the paper makes IPC analysis, inventor analysis, company analysis, country analysis and overall trend analysis. After analysis on relevant term segmentation and word segmentation of component words, technical terms and functional terms of the traditional and simplified Chinese and English patents, the paper establishes the technical/functional word cluster, and produces the first and second technical hierarchies of technical/functional matrices. From the first and second technical hierarchies of technical/functional matrices, the paper establishes the first and second technical hierarchies of different item techniques and companies, analysis chart of their core competitiveness, competitor R&D activity trend chart, and technical-hierarchy activity trend chart. Finally, the paper practically verifies whether relevant patents belong to the first or second technical hierarchy, and proves the feasibility of this research approach. Relevant traditional and simplified Chinese and English patents of LED reading lamp searched above, and the results of IPC analysis and technical analysis mentioned above can be provided as a reference for enterprises and engineers to save their R&D time in production of new patients, and for enterprises to know the core competitiveness of their competitors.

參考文獻
【1】「國際專利分類表第七版使用指南」，經濟部智慧財產局，(2001)。
【2】「國際專利分類之研究」，經濟部中央標準局。
【3】 Overview of the Classification System,
http://www.uspto.gov/main/patents.htm美國專利商標局（USPTO）。
【4】The EPO guide to patent information on the internet,
http://epart.epo.org/dwl/espacenet_manual.pdf歐洲專利組織(EPO)
【5】Trappey, C. V., Trappey, A. J. C., and Wu, C. Y., “ Clustering patents using non-exhaustive overlaps,” Journal of Systems Science and Systems Engineering, Vol.19,No.2,pp.162-181,(2010).
【6】Trappey, C. V., Trappey, A. J. C., and Wu, C. Y., “ Automatic patent document summarization for collaborative knowledge systems and services,” Journal of Systems Science and Systems Engineering, Vol.18,No.1,pp.71-94,(2009).
【7】Singhal, A. and Salton, G., Automatic Text Browsing Using Vector Space Model, Technical Report, Department of Computer Science, Cornell University, (1993).
【8】Lawrence, S. and Giles, C. L., “Searching the World Wilde Web General and Scientific Information Access,” IEEE Communication, Vol.37,pp.116-122, (1999).
【9】陳哲宏、陳逸南、謝銘洋、徐宏昇，「專利法解讀」，月旦出版公司，第14-15頁，(1994)。
【10】Gruber, T., “Ontolingua: A translation approach to portable ontology specifications ,” Knowledge Acquisition, pp.199-200(1993).
【11】Oleary, D. E., “Enterprise knowledge management,” IEEE Computer, Vol.31, pp.54-61 (1998).
【12】杜家宏，「整合繁體和簡體中文及英文斷詞斷字系統之發光二極體檯燈專利分析研究」，碩士論文，國立台灣科技大學機械工程學系，台北,(2011)。
【13】吳俞灃，「LED水族燈之相關專利分析研究」，碩士論文，國立台灣科技大學機械工程學系，台北(2012)
【14】Salton, G. and McGill, M., Introduction to Modern Information Retrieval, McGraw-Hill,New York, (1983).
【15】曾元顯，「圖書館學與資訊科學大辭典」，國家教育研究院，(2011)
【16】Vallet, D., Fernandez, M., and Castells, P., “An Adaptation of the Vector-Space Model for Ontology-Based Information Retrieval,” IEEE Knowledge and Data Engineering,Vol.19, No.2, pp.261-272, (2005).
【17】Wang, S., Sun, J., Gao, B. J., and Ma, J., “Adapting Vector Space Model to Ranking-based Collaborative Filtering,” CIKM '12 Proceedings of the 21st ACM international conference on Information and knowledge management , pp. 1487-1491, New York , USA, (2012) .
【18】jie, G. and Chen, L. C., “Research of improved IF-IDF Weighting algorithm,” International Conference on Information Science and Engineering (ICISE) , pp.2304-2307, Hangzhou ,China,(2010).
【19】Zhang, G. H. and Odbal, “Sentence Alignment For Web Page Text Based On Vector Space Model,” International Conference on Computer Science and Information Processing(CSIP),pp.167-170, Shaanxi ,China,(2012).
【20】Ryan, J. M., Roshan, J., and Song, L., “Integrated Thermal Management Techniques for High Power Electronic Devices,” Applied Thermal Engineering,Vol.24, No.8-9, pp.1143-1156, (2004).
【21】莊育宗，「可攜式閱讀燈具」，台灣發明專利，M376685, (2010)。
【22】李克勤，「可攜帶型的USB-LED閱讀燈」，台灣發明專利，D145568, (2012)。
【23】Tai, C. Y., “Lighting system utilizing cylindrical lens and fluorescent lamps designed for specific use in aquarium and desk lighting applications,” U.S. Patent Application Publication, US2003/0025457, (2003).
【24】Seiya Company 國際品牌大廠代工製造商之供應施工夥伴,
http://www.ledcase.com/light-2.htm
【25】林武憲，「LED整合中英文斷詞斷字系統之專利分析」，碩士論文，國立台灣科技大學機械工程學系，台北,(2010)。