以文字探勘與偏最小平方結構方程模型挖掘影響消費者接受智能理財之關鍵要素

Abstract

近年來，科技創新顛覆各個金融領域，數位金融的改變、行動支付的使用、機器人理財日益普及、人工智慧的廣為應用，皆改變傳統金融產業的思維。在財富管理中，投資理財的工具也不例外。其中，最為廣泛討論的議題，為『智能投資』。傳統理財服務，乃透過理專或投資顧問進行，智能投資的出現，提供新的訊息與方法。智能投資是目前國際金融的趨勢，如何穩健獲利並控制風險，為多數投資人追求之渴望境界。而智能投資為新型投資理財工具，一般投資人對其瞭解較少，調查接受意願較為不易。而由於概念新穎，相關研究亦少。為跨越此研究缺口，調查影響投資大眾接受智能理財的因素，本研究擬利用文字探勘技術，擷取社群網站中，有關智能投資的相關貼文，並使用主題分析(Topic Modeling)，萃取貼文中之主題，作為影響投資者使用智能理財意願之關鍵要素。之後依據技術接受模式(Technology Acceptance Model，TAM)提出假設，並將各關鍵要素歸入各主題中，再以偏最小平方結構方程(PLS-SEM)驗證各主題之關聯關係是否顯著。本研究以探勘 Twitter.com 驗證分析架構之可行性。依據實證研究結果，影響投資人使用智能理財的主要因素為感知智能理財之易用性。通過提高感知易用性，消費者使用智能理財投資工具之意願，將得到提昇。本研究所定義之理論模型與驗證完善之分析架構，未來可用於探勘影響投資者接受其他新型金融科技之關案要素，也可作為金融業發展行銷策略與建置服務之參考。

In recent years, technological innovation has changed various segmentations of the financial sector. The changes in digital finance, the use of mobile payments, the increasing popularity of robotic financial management, and the widespread application of artificial intelligence have all changed the thinking of the traditional financial industry. In wealth management, investment and financial management tools are no exception. Among them, the most widely discussed topic is"Robo-advisors". Traditional wealth management services are carried out through specialists or investment consultants. The emergence of Robo-advisors provides new information and methods. Robo-advisors are the current trend of the world. How to make steady profits and control risks is the ideal state pursued by most investors. Robo-advisor is a new investment and financial management tool. However, most investors know little about it, and it is not easy to investigate the willingness for accepting such technique. Moreover, very few researches try to bridge this research gap and investigate the factors that influence the investors’ acceptance of Robo-advisors. Thus, this research intends to use text mining techniques to extract relevant posts on Robo-advisors from social networking sites, and use the Latent Dirichelet Allocation (LDA) to extract major topics. The topics being retrieved from the articles will serve as key factors that will affects investors' willingness to adopt Robo-advisors After that, hypotheses were defined based on the Technology Acceptance Model (TAM). The key elements derived by LDA will be grouped into each topic. Then, partial least square structural equation modeling (PLS-SEM) technique will be used to test whether the correlation between each pair of topics is significant. This study is to mine Twitter.com to verify the feasibility of the analytic framework. Based on the results of empirical research, the main factor affecting investors' use of Robo-advisors is the perceived ease of use. By improving perceived ease of use, investors' willingness to adopt Robo-advisors will be enhanced. The theoretic model and well-validated analytic framework defined by this research can be used in the future to explore factors influencing investors' acceptance of other new financial technology. The analytic framework can also be used as a basis for developing marketing strategies and develop novel financial services.