一种大容量文本集的智能检索方法

摘要

分析了潜在语义模型,研究了潜在语义空间中文本的表示方法,提出了一种大容量文本集的检索策略.检索过程由粗粒度非相关剔除和相关文本的精确检索两个步骤组成.使用潜在语义空间模型对文本集进行初步的筛选,剔除非相关文本;使用大规模文本检索方法对相关文本在段落一级进行精确检索,其中为了提高检索的执行效率,在检索算法中引入了遗传算法;输出这些候选的段落序号.实验结果证明了这种方法的有效性和高效性.%An information retrieval approach for large-scale collections of full-text document is proposed according to latent model analysis and investigation of latent space-based text representation form. The retrieval process is divided into rough irrelative full-text documents culling procedure,and relative full-text document precise searching prueedure.lrrelative documents are removed by the first procedure. Relative full-text documents are retrieved in passage level by the second one,and in this process, GA algorithm is introduced in order to achieve best performance. Finally, the candidate passage indices are returned.The validity and high efficiency of the proposed method are shown by experimental results.