Space efficient caching of query results in search engines

Abstract

Web search engines serve millions of query requests per day. Caching query results is one of the most crucial mechanisms to cope with such a demanding load. In this paper, we propose an efficient storage model to cache document identifiers of query results. Essentially, we first cluster queries that have common result documents. Next, for each cluster, we attempt to store those common document identifiers in a more compact manner. Experimental results reveal that the proposed storage model achieves space reduction of up to 4%. The proposed model is envisioned to improve the cache hit rate and system throughput as it allows storing more query results within a particular cache space, in return to a negligible increase in the cost of preparing the final query result page. © 2008 IEEE.