一种快速的随机分块文本检索算法  

A Fast Text Retrieval Algorithm based on Random Blocking

作　　者：赵亚慧[1] 刘金广[2] 崔荣一[1] 

机构地区：[1]延边大学工学院计算机科学与技术学科智能信息处理研究室,延吉133002 [2]黄淮学院计算机科学系,驻马店463000

出　　处：《武汉理工大学学报》2010年第23期135-139,共5页Journal of Wuhan University of Technology

基　　金：吉林省教育厅资助项目(吉教科合字2009第7号)

摘　　要：针对向量空间模型(VSM)在信息检索中存在脱离上下文语境的缺陷,提出一种新的随机分块文本检索算法。该算法利用潜在语义分析技术在低维空间中对检索文本与待检索文本的各个段落进行投影,使用SVD技术对其进行了降维处理,得到文本特征,然后依据该文本特征进行相似的匹配,获得检索结果。由于充分考虑了文本的上下文语义信息,算法能够有效提高检索效率。实验结果表明,该算法在检索时间、查准率、召回率及综合性能等方面均得到较好的效果。A new text retrieval algorithm based on random blocking was proposed to overcome the drawback of VSM which doesn＇t make use of the text-context in information retrieval.Firstly,latent semantic analysis（LSA） was employed to project the paragraphs of retrieve texts and to be retrieval texts into lower-dimensional space,and then processing of dimension reduction was adopted to obtain text features by SVD.Then,similarity matching of the text features was used to achieve retrieval results.Due to the information of text-context was took into fully consider,the proposed algorithm improved retrieval effectively.The experiment results demonstrate that the proposed method has high performance in retrieval time,precision rate and combination property etc.

关 键 词：随机分块检索算法 向量空间模型 SVD技术 文本相似度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]