請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74235
標題: | 基於軌跡的協同過濾方法支持企業技術發展預測 A Trajectory Based Collaborative Filtering Approach to Support Firm Technological Development Forecasting |
作者: | Wen-Heng Qiu 丘文恒 |
指導教授: | 魏志平(Chih-Ping Wei) |
關鍵字: | 技術軌跡,研發,協同過濾,技術發展預測, technological trajectory,R&D,collaborative filtering,technological development forecasting, |
出版年 : | 2019 |
學位: | 碩士 |
摘要: | 技術競爭被認為是數字時代競爭和創新戰略的關鍵維度。技術發展中最重要 的任務之一是在考慮到公司目前的限制因素(例如研發成本)下確定研發方向。
預測公司自身或競爭對手的技術發展,可以提高公司的研發實用性和表現, 同時降低研發投資風險水平。結合企業常規研發計畫和技術發展預測,企業可以 盡量避免研究與開發的盲點。企業還可以利用技術發展預測來尋找潛在技術互補 領域的企業,以最大化合併或收購的效益。此外,它還可以預先通過取代替代發 明(圍欄策略)以阻止競爭對手的商業活動,並避免技術被其他競爭者所擁有的 風險,或作為訴訟和交叉許可(遊戲策略)討價還價的籌碼。 本文提出了一種旨在支持公司技術發展預測的基於技術軌跡的協同過濾算 法,並獲得優於性能基準的表現。研究主要將公司的研究順序考慮在內,並嘗試 探索不同相似度計算方法及其對應的推薦質量。這種方法僅需使用公開的專利數 據,便可以簡單易行的方式預測公司自身或競爭對手的研究方向。因而具有易於 實作,低成本的特點,並可作為傳統的基於文本挖掘方法的補充。 Technological rivalry is recognized as a key dimension of competition and innovation strategies in the digital era. One of the most important tasks in the technological development is determining the R&D directions, taking into consideration the firms’ present constraints, such as the costs of R&D. Forecasting a firm’s own or competitors’ technological development, can improve a firm’s R&D practicality and performance while reducing the level of R&D investment risk. Moreover, combining original scientific effort and technological development forecasting, a firm can try the best to avoid blind spot of research works. A firm can also take the advantage of technological development forecasting to merge or acquire other companies that are active in complementary technological fields. In addition, it can also block the commercial endeavors of rivals and by preempting substitute inventions (fence strategy), to avoid the risk of hold-up by other technology owners, or as a bargaining chip in litigation and cross-licensing (play strategy). In this paper, we present a technological trajectory-based collaborative filtering algorithm to support firm technological development forecasting, which outperforms the performance benchmarks. We take a firm’s research order into account and attempt to explore different values of Firm-IPC table and its corresponding quality of recommendations. This approach can give a simple and easy-to-implement way to forecast a firm’s own or competitor’s research orientation by only using publicly available patent data. It is also an efficient, low-cost way for complementing the traditional text mining-based approach. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74235 |
DOI: | 10.6342/NTU201903093 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊管理學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-108-1.pdf 目前未授權公開取用 | 1.58 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。