Novel Approaches for Integrating MART1 Clustering Based Pre-Fetching Technique with Web Caching

Introduction

Web pre-fetching used to improve the performance of web based information retrieval system by pre-fetching web pages into the cache before actual request arrives. Therefore, Web pre-fetching involves following steps

• •

  Prediction system for anticipating user future request based on their previous access pattern.

• •

  Pre-fetching and loading them into the cache.

Thus, web pre-fetching also involves web caching. However, the web caching and web pre-fetching were addressed separately by many researchers in the past. Only few research works were carried out in integration of web caching and web pre-fetching. The following section gives an overview of all such a techniques.

Most of the existing pre-fetching techniques employ single object pre-fetching technique, which is handled by traditional cache replacement policies. However, in the clustering based pre-fetching technique, multiple objects are pre-fetched. Hence existing replacement algorithms are not suitable for multiple object pre-fetching technique. Hence this chapter provides different heuristics techniques in cache replacement policy which will ultimately reduces the number of objects pre-fetched, increases prediction accuracy and hence saves the bandwidth.

Web Caching Policies

Web caching policy decides which pages to be removed from the cache when there is no enough space for storing new page. The main goals of web caching are

• •

  It has to use the cache space more efficiently,

• •

  It has to optimize one or several web cache performance metrics such as cache hit rate, byte hit rate, response time and so on.

• •

  It has to improve CPU and memory utilization.