Pre-eminence of Combined Web Pre-fetching and Web Caching Based on Machine Learning Technique

It is always wise to keep the frequently used information by various users in the cache memory. This makes the users feel that their needed information is available almost immediately. Apart from experiencing less access delay, caching process also helps in better bandwidth utilization and load reduction in the origin server. Pre-fetching is the process of fetching few of the Web pages in advance by assuming that those pages will be needed by the user in the near future. Combining pre-fetching and caching techniques results in experiencing much less access delay and much better bandwidth utilization. Many works have been reported in the literature, separately for Web caching techniques and pre-fetching of Web pages. In this paper, pre-fetching technique that uses clustering is combined with SVM (support vector machine)-LRU algorithm, a machine learning method for Web proxy caching. By using real-time data, it is demonstrated that the latter approach will be advantageous than clustering-based pre-fetching technique using traditional LRU-based caching policy. The efficiency of our proposed method is also compared with caching using Bayesian networks and neuro-fuzzy system.