近年來網路蓬勃發展對於多數人而言，上網已不再是資訊人員的專利，而成為了日常生活中不可或缺的一部分。推薦系統可以根據使用者個人的偏好或需求，協助使用者從大量的選擇中，找到使用者感興趣或相關性高的資訊，顯著的幫助人們解決此一問題。因此自從電子商務系統問世以來，推薦系統一直是研究的重點。
情境感知屬性(contextual Aware)是能夠將使用者所需的資訊
，透過感應器或無線網路的協助及依據當時的情境因素，提供適當資訊。主要包含時間、天氣、地點、距離等因素，近年來越來越多推薦系統使用情境因素。
本研究利用協同過濾的方式增加評分數量來解決稀疏性的問題。並使用K-means分群，接著利用決策樹(decision tree)替有相同喜好的同一群建立預測模型，本研究透過決策樹了解每個群對距離的重視程度。最後經過實驗，可以發現對大多數人而言距離為重要屬性，並且在推薦系統中加入距離可以提高準確率。
In recent years, with the rapid development of the Internet, logging onto the Internet is indispensable not only for information administrators but also for individuals. Based on user preferences and needs, recommendation systems select from massive information to help users find interested or highly related information, thus significantly improving precision of recommendation for users. Therefore, since the emergence of e-commerce systems, recommendation systems have been focus of research.
Contextual aware attribute is to change user requested information to appropriate information by the assistance of the sensors or wireless network based on contextual factors which mainly consist of time, weather, location, distance, etc. In recent years, more and more recommendation systems use contextual factors.
In this study, we use collaborative filtering mechanisms to increase the number of ratings to solve the sparsity problem and use the K-means and decision tree for clustering and to create the prediction model for groups with the same preferences. By using decision trees, we can know the importance level of distance in each cluster. To conclude, for most of people, the distance is an important attribute, and adding distance attribute into recommendation system can indeed improve the precision of recommendations for users.
