摘要： 根据文本向量的维分布的稀疏性,提出了基于倒排索引的文本相似搜索算法。该算法通过倒排索引进行维过滤,快速获得尽量小的目标集的超集。实验表明,在海量文本环境下,该方法虽略微损失准确度,但其速度远远高于传统基于多维索引的算法。

关键词: 相似搜索, 倒排索引, 多维索引, 文本相似搜索

Abstract: For the dimensions sparseness of the text set, a new similarity search algorithm for text set is proposed, which is based on inverted-index. The algorithm can quickly gain a super-set for the targets by search on inverted-index. Experiments show that the algorithm is faster than the algorithm based on multi-dimension index for huge text set, while a little nicetylosing.

中图分类号: 

• TP391.3