提出了一种基于语句的查询扩展方法以及语句向量的融合策略,使得扩展后的查询语句的查询性能优于原始查询语句;基于微软高性能计算平台HPC Server和查询扩展策略,设计实现了一个分布式文本检索系统DQSSQE.实验结果表明,在检索性能方面,所提出的查询扩展策略能够有效的提高查准率,召回率上也有一定的提高;在分布式检索计算性能方面,DQSSQE系统具有较好的计算加速比,随着文本集规模的增加,其计算性能的优越性体现明显.
In this paper, a query expansion method based on sentences and a sentence vectors combination strategy are proposed to improve the query performance. A distributed text retrieval system DQSSQE is designed based on Mi- crosoft HPC Server platform and query expansion strategy. The experiment result shows that the proposed query ex- pansion strategy improves the precision ration greatly, and improves the recall ratio as well. At the same time, DQSSQE system gets a higher computation speedup ratio, and the more large the text set is, the higher performance the system will get, compared to the ordinary text retrieval systems.