應用文件探勘於專利文件之技術分析－以磁阻性隨機存取記憶體為例

Abstract

專利文件包含了豐富的技術資訊，公司能透過專利資訊的分析，瞭解某項技術或產品的發展現況。一項技術通常係由許多相關的技術類別所組成，過往多以該技術領域的專家根據專利分類號判斷其所包含的技術類別，然以專家進行技術的分類，不僅秏時且秏成本，且可能會有分類一致性的問題。因此要如何以快速及客觀的方法來取代或輔助人工的分類，找出一項技術所包含的技術類別，為相當重要的議題。文件探勘為近年來興起的新技術，其能從大量的文件中，挖掘隱含且有用的知識，然卻鮮少應用於專利文件上，因此本研究將嘗試以文件探勘的技術，以磁阻性隨機存取記憶體(MRAM)為例，根據相同專利分類號的專利文件，彼此在技術內容的相似度，找出MRAM所包含的技術類別，並以CHI研究公司所提出的專利指標，對MRAM的相關技術類別做進一步的分析。

Patent documents not only provide the technological information, but also reflect the technological development and trend. Companies can discover the technological development after conducting patent analysis. One technology usually comprises several technological fields, so the prior work at the time of performing a patent analysis is to confirm the technological fields of one technology. In the past, the works to identify the technological fields is often done by experts. Experts make the judgment of the comprised technological fields based on the International Patent Classification used by patent databases. However, this is a time-consuming and costly job and may cause inconsistent problems. In order to solve the problems caused by manual judgment, we introduce an emerging method, called text mining, to clarify the technological fields. The text mining method can discovery the implicit and useful knowledge from a large number of documents. Patents documents related to Magnetic Random Access Memory are selected as an example to demonstrate how to use text mining in clarifying technological fields. The technological fields are classified based on the similarity between patent documents. After classification, patent performances of each technological field is analyzed and compared according to the commonly used patent indicators proposed by CHI Research Corporation. At last we provide suggestions for future research.