Research on the second-level query results cache in Web search engines

摘要

Query results cache can improve the efficiency of Web search engines significantly. In this paper, we use SOGOU search engine query logs to study the characteristic of the second-level query results cache's workload and the impact of different cache replacement policies. Experimental results show that static cache without any policy can achieve the optimal cache hit ratio. When the capacity of the second-level query results cache becomes larger, LIRS achieves the optimal cache hit ratio, and MQ is the optimal replacement policy in the case of the largest capacity.