運用文字探勘建置專利技術功效矩陣之成效評估

Evaluation of the Application Text Mining to Build Patent Technology Function Matrix

10.6270/JIBM.2013.4(2)163

黃嘉彥(Jia-Yen Huang)；楊竣徨(Chun-Huang Yang)

專利地圖 ； 文字探勘 ； 特徵值選取 ； 技術功效矩陣圖 ； Patent Map ； Text Mining ； Feature Extraction ； TF-IDF ； Technology-Function Matrix

創新與經營管理學刊

4卷2期（2013 / 12 / 01）

162 - 176

繁體中文

在知識經濟時代，專利地圖分析是瞭解產業發展趨勢與競爭對手研發策略的必要工具。其中，透過技術功效矩陣除可解析對手佈局策略，亦可從中找尋創新研發的機會，以及迴避專利地雷。而拜資訊科技進步之賜，參考資訊取得容易，因此整體專利數量每年急速增加。然而欲從龐大的專利文件中建構一完整之技術功效矩陣圖，需耗費許多人力時間，因而增加企業支出成本。本研究主要目的為利用文字探勘（Text Mining），配合TF-IDF特徵值選取方法，分析專利之技術與功效，以提高建構技術功效矩陣圖之效率。經與人工方式建構之技術功效矩陣圖之結果與時效比較，本研究以文字探勘搭配TF-IDF建構技術功效矩陣圖的方式，準確率高達80%，並可有效減少以人工方式所需耗費之時間。本研究以居家常見之紗窗為範例，證實以文字探勘輔助建置技術功效矩陣的方法，將可降低企業成本支出，進而提升企業於產業中的競爭優勢。

In the age of knowledge economy, patent map analysis is a necessary tool for understanding industry trends and R&D strategies of competitors. Technical function matrix, one of important analysis schemes of patent map, can not only be used to reveal the layout strategy of opponents, but also to discover new opportunities for innovative R&D and to avoid patent infringement. Modern companies generally apply for patent as a means for protecting their own intellectual properties and competitiveness strength. Thanks to the advancement of information technology, the overall number of patents grows rapidly. It takes huge of manpower to build complete technology function matrix, thus increasing the cost of business spending. This study aims to use text mining, with the aid of TF-IDF eigenvalues selection methods, to analyze the techniques and functions of patents so as to improve the efficiency of constructing technology function matrix. Comparing with the manually made technical function matrix, the text-mining based technique can achieve correctness up to 80%, and can effectively reduce the construction time. Using home screens as an example, this study has confirmed that the text-mining based technology function matrix method is able to reduce costs, and thus enhance the enterprise's competitiveness.

1. 葉子諒(2017)。電腦輔助專利創新性評估與檢索系統。清華大學工業工程與工程管理學系學位論文。2017。1-96。