Research on the Search Engine Cache RepIacement Strategy Based on MuItipIe Query Attributes
Cache is a very important technology in search engine, which can significantly save query computation processing, improve query re-sponse and improve system throughput, which are widely applied by the academia and the industry. Current cache replacement policy does not take full advantage of search engine queries of multiple access feature information, does not take advantage of query distribution, also deficiencies exist in the traditional replacement policy when used in search engines. For the above problems, studies query distribu-tion features, analyses the insufficient of existing cache replace strategies, then proposes integrated value function model represent query future heat value based on query access features, analyses search engine query log for fine grain degrees, gets each query's daily access characteristics of detailed records, and based on multiple return analysis in the minimum II multiplication calculation to get the unknown parameter in the function model, designs cache management policy integrate current dynamic access attributes with the heat value of the query in the future, hit ratio test of replace management strategy through real query shows that, in contrast with traditional cache replace-ment strategy, this replacement strategy significantly exceeds them in hit rate.
NETL
NSTL
万方数据
维普
