全球資訊網是網際網路中最為廣泛使用於傳播、獲取訊息的服務。不幸的是，用戶經常會因網路擁塞造成長時間的訪問延遲。網頁緩存預取是一種常見用於降低用戶回應時間的方法。在本文中，我們建立了基於推薦系統的快取代理伺服器(recommendation based cache proxy for web browsers)，利用推薦系統預測將會被下載的網頁。此系統將被預測網頁的統一資源定位位址和網頁元件預取至快取中，減少使用者瀏覽網頁的網頁讀取時間。利用真實的數據集，我們比較了項目為基準的協同過濾、基於流行推薦、隨機推薦，並證明了使用以項目為基準的協同過濾可令系統達到最佳的效能。另外和靜態的快取策略，先進先出演算法、最久未使用演算法等做比較。
;The World Wide Web is the most widely used application for information access and dissemination on the Internet. Unfortunately, users often experience long access latency due to network congestion. Web caching prefetching is a common approach used to reduce the response time perceived by users. In this study, we build a recommendation based cache proxy for web browsers, which use the recommendation system to predict web pages to be downloaded. The system then prefetches the URLs and objects into the cache to reduce the web load time when the client visits the web pages. Using true datasets, we show that the recommendation based cache proxy using IBCF achieves the best performance in a variety of metrics compared with the system based on POPULAR and RANDOM as well as the static cache strategies, such as FIFO and LRU.
