基于潜在语义分析的构件聚类改进方法  

Improved Method of Component Clustering Based on Latent Semantic Analysis

作　　者：任姚鹏[1,2] 陈立潮[1] 张英俊[1] 谢斌红[1] 

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024 [2]运城学院计算机科学与技术系,山西运城044000

出　　处：《计算机工程》2011年第4期67-69,共3页Computer Engineering

基　　金：山西省自然科学基金资助项目"基于语义的构件检索关键技术研究"(2009011022-1)

摘　　要：针对基于向量空间模型的构件聚类方法存在高维稀疏、无法解决同义词等问题,采用基于潜在语义分析模型对构件进行聚类分析。从用户关注点出发,通过引入等级策略提出一种基于潜在语义分析的构件聚类改进算法。实验结果表明,该方法能够提高构件聚类质量,使构件聚类结果更符合用户需求和更加人性化,提高构件检索效率和准确性。Aiming at the problem that the current component clustering method based on Vector Space ModeI（VSM） has the high-dimensional sparse and is unable to solve the synonym, Latent Semantic Analysis（LSA） model is used to cluster the components, meanwhile from the point of user attention, grade strategy is introduced. An improved method of component clustering is proposed based on LSA model. The method is proved effective by experiments, which can improve the quality of component clustering and make the component clustering result better serve user requirement and even more humanized. It can promote the efficiency and accuracy of component retrieval.

关 键 词：刻面分类 潜在语义分析 等级策略 构件聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]