透過您的圖書館登入
IP:18.219.236.199
  • 學位論文

運用專利搜尋與文字探勘來探討產品設計之演化趨勢模式

Applying Patent Search and Text Mining to Explore the Evolutionary Trend Modeling for Product Design

指導教授 : 劉天倫

摘要


在使用TRIZ理論進行產品設計時往往需要選擇適合工具解決問題,過去大多是以人為的判斷為主,而產品開發流程以提升產品設計理想性(Ideality)為目標,來達到跳躍式的創新。藉由TRIZ中演化趨勢的演化漸進線可以有效的提升理想性,然而選擇適合的演化漸進線來分析產品的內容是相當關鍵的。因此本論文運用文字探勘及語義相似度計算來分析專利文件中可能有關的演化趨勢,進而協助產品開發者選擇適合解決設計中問題的演化漸進線。本研究首先建立了37演化趨勢SAO(Subject-Action-Object)語法結構關鍵字詞集,並參考了與產品設計需求相關的研究來選定本論文之實驗題目-步矩橢圓機以及建立出SAO(Subject-Action-Object)語法結構之產品設計需求,再從專利資料庫擷取步矩橢圓機之專利文本,利用文字探勘對專利文本進行依存分析(Dependency Parsing)來擷取出SAO語法結構,通過Word2Vec訓練的詞向量進行語義相似度計算來分類該專利文件符合的演化趨勢以及設計需求,最後將結合演化趨勢分類結果和設計需求分類結果來探討步矩橢圓機的產品設計在演化趨勢應用上的推薦。

並列摘要


When applying the TRIZ theory for product development processes, it is often necessary to choose suitable tools to solve problems. In the past, most of them were based on human judgment, and the product development process aimed to improve the ideality of product design to achieve leapfrog innovation. The ideality can be effectively improved by the trend lines of the evolution trends in TRIZ, but it is quite critical to choose a suitable evolutionary lines to analyze the content of the product. Therefore, this paper uses text mining to analyze the evolutionary trends that may be relevant in patent documents, and then assists product developers to choose an evolutionary lines suitable for solving design problems. This research first constructed the 37 evolutionary trends SAO (Subject-Action-Object) grammatical structure keywords set, and referred to the papers related to product design requirements to select the experimental topic of this research - elliptical machine and established SAO grammatical structure of product design requirements, and then extract the patents texts of the elliptical machine from the patent database, and use text mining to perform dependency parsing on the patent text to extract SAO grammatical structure. The semantic similarity calculation of the word vector trained by Word2Vec is used to classify the evolutionary trends and design requirements of the patent documents. Finally, the product design of the elliptical machine will be discussed in combination with the evolutionary trends classification results and the design requirements classification results recommendations on evolutionary trends applications.

參考文獻


一、 中文參考文獻
1. 王志超,「應用QFD與 TRIZ 在新產品設計之研究 ‐以自由步距橢圓機為例」,逢甲大學工業工程與系統管理學系碩士在職專班碩士論文,2013。
2. 宁建飞、刘降珍,「融合Word2vec 与 TextRank的关键词抽取研究」,现代图书情报技术,2016,P20 – 27,2016。
3. 李晓、解辉、李立杰,「基于Word2vec的句子语义相似度计算研究」,计算机科学, 44(9):256 – 260,2017。
4. 李彥璇,「基於類神經網路之繁體中文依存句法分析器」,國立交通大學電信工程研究所碩士班碩士論文,2019。

延伸閱讀