蒐集使用者行為資訊以提供個人化的服務與提升服務品質已日漸重要,因此如何準確地獲取使用者於服務使用時的行為便成為重要的議題。目前在學術研究上已經有許多關於使用者行為獲取的機制與相關研究,其中最常見的作法為以圖像辨識方式獲取使用者行為,但其消耗CPU效能,且同時在圖像劇烈變化之下無法正確地辨識以及獲取使用者行為。目前多數相關研究的缺點是僅針對固定視窗大小進行辨識,且在多數常用的功能皆已提供快捷鍵情況下,目前研究並無提供辨識快捷鍵的功能。為改善這些缺點,本研究提出視窗環境下之使用者行為獲取機制研究,將應用程式的視窗以一個網格48x48的方式切割,並設定好每一個網格上代表的元件,以獲取使用者在應用程式上的行為。同時本研究亦提供自動重算視窗網格的改變,能即時獲取使用者行為。我們已在Windows作業系統上完成系統雛型,同時也進行了兩個實驗。實驗結果顯示本研究能正確地針對Windows以及GTK所開發的視窗程式進行辨識並獲取使用者行為。;With a collection of user behavior information to provide a personalized service and to improve the service quality has become increasingly important. How to accurately obtain user behavior when user using the service or application has become an important issue. Currently, in the academic study has a number of mechanisms related research on user behavior acquired. One of the most famous practice was image recognition for acquisition user behavior, but its consumption performance, and at the same time it can’t properly identify and obtain user behavior under the picture dramatically. In the majority of studies conducted to identify only for a fixed window size. On the other hand, the most commonly used functions have provide a hotkeys to greatly enhance user interaction with the GUI interface, but these studies didn’t provide identification and obtain what function is the hotkeys was press. To improve these shortcomings, our study presents a user behavior acquire mechanisms under windows program environment, to recognize the whole windows program to get the user behavior on the program’s window. At the same time, our study also provide an automatic recalculation adjusted of this specified location for the changes of the windows program location, so that it’s possible to automatically identify and capture the user behavior when user change the windows program size and click on the specified location or specified button. We have completed a prototype system on Windows Operating System, also carried out two experiments. The results show that can correctly against Windows and GTK programs identify and obtain user behavior.