基於文字探勘的專利分析與預測

Text Mining Based Patent Analysis and Forecasting

邵子安

指導教授 ： 張適宇

國立清華大學/電機資訊學院/通訊工程研究所/碩士(2012年)

摘要

MIMO (multiple-input and multiple-output) 技術在無線通訊領域備受關注，因為它提供了在不增加發射功率或額外頻寬的情況下，增加鏈接範圍跟數據吞吐量。可以經由在天線上散佈相同的總功率，達到陣列增益以增加頻譜效率，或是達到分集增益以增加鏈路的可靠度。因為這些特性，MIMO成為了現代無線通訊標準(IEEE 802.11n (Wi-Fi), 4G, 3GPP Long Term Evolution, WiMAX (Worldwide Interoperability for Microwave Access) 和 HSPA+)。雖然MIMO是一種現代的無線通信標準的重要組成部分，但仍然沒有人對MIMO技術進行預測或專利分析研究。專利分析是一個很好的方式來處理競爭分析與技術趨勢分析，目前有很多文章使用專利文獻進行競爭分析與技術趨勢分析，因為專利是科技知識與商業知識中重要的資源。研發人員、專家和研究人員可以利用專利分析的結果，規劃戰略評估的競爭地位，以避免侵權。本研究首先提出了一個羅吉斯回歸演算法預測MIMO技術的趨勢，然後通過使用文字探勘演算法推導出MIMO技術的專利地圖。最後提出蟻群優化（ACO）算法解決最小生成樹（MST）的問題。擴展出的MIMO能力集合可以用來幫助一間公司達到成功營銷。

關鍵字

文字探勘 ； 專利預測 ； 專利分析

並列摘要

MIMO (multiple-input and multiple-output) technology has attracted attention in wireless communications because it provides significant increases in link range and data throughput without increasing transmission power or additional bandwidth. The goal can be achieved by spreading the same total transmission power over the antennas to reach an array gain that improves the spectral efficiency or to reach a diversity gain that improves the link reliability. MIMO is an important part of modern wireless communication standards (IEEE 802.11n (Wi-Fi), 4G, 3GPP Long Term Evolution, WiMAX (Worldwide Interoperability for Microwave Access) and HSPA+) because of these properties. Although MIMO is an important part of modern wireless communication standards, still no one has researching into MIMO technology forecasting or MIMO patent analysis. Patent analysis is a good way in competitive analysis and technology trend analysis. Lots of paper using patent information in competitive analysis and technology trend analysis because patents are an ample source of technical and commercial knowledge. R&D practitioners, specialists and researchers can take the results of patent analysis to plan strategies to assess competitive position and to avoid infringement. This research first proposes a logistic regression algorithm to forecast the trend of the MIMO technology and then derive patent maps of the MIMO technology by using the text mining algorithm. Finally ant colony optimization (ACO) algorithm will be proposed to solve minimum spanning tree (MST) problem. The competence set will be expanded for a MIMO design house to help a firm for achieving marketing success.

參考文獻

[1] Liu, S.J. and Shyu, J., "Strategic planning for technology development with patent analysis", International Journal of Technology Management 13, 5 (1997), pp. 661--680.

[2] Abraham, B.P. and Moitra, S.D., "Innovation assessment through patent analysis", Technovation 21, 4 (2001), pp. 245--252.

[3] Hicks, D. and Breitzman, T. and Olivastro, D. and Hamilton, K., "The changing composition of innovative activity in the US--a portrait based on patent analysis", Research policy 30, 4 (2001), pp. 681--703.

[4] Daim, T.U. and Rueda, G. and Martin, H. and Gerdsri, P., "Forecasting emerging technologies: Use of bibliometrics and patent analysis", Technological Forecasting and Social Change 73, 8 (2006), pp. 981--1012.

[5] Trappey, C.V. and Wu, H.Y. and Taghaboni-Dutta, F. and Trappey, A.J.C., "Using patent data for technology forecasting: China RFID patent analysis", Advanced Engineering Informatics 25, 1 (2011), pp. 53--64.