| 基于奇异值分解的关系相似度计算方法 |
| |
| 引用本文： | 郭丽,刘磊,季铎,蔡东风.基于奇异值分解的关系相似度计算方法[J].沈阳航空工业学院学报,2009,26(3):78-81. |
| |
| 作者姓名： | 郭丽 刘磊 季铎 蔡东风 |
| |
| 作者单位： | 沈阳航空工业学院知识工程中心,辽宁,沈阳,110034 |
| |
| 摘 要： | 关系相似度计算在信息检索等领域有重要的应用.目前,关系相似度主要通过向量空间模型计算,但数据稀疏严重影响计算效果.为此提出以上下文模板表示特征,并通过对特征矩阵的奇异值分解降维消除噪声,冉通过向量余弦夹角计算关系相似度.实验表明,该算法比传统的向量空间模型算法在分类准确率方面有明显提高.
|
| 关 键 词： | 关系相似度 向量空间模型 奇异值分解 |
Calculation method of relational similarity based on singular value decomposition |
| |
| GUO Li,LIU Lei,JI Duo,CAI Dong-feng.Calculation method of relational similarity based on singular value decomposition[J].Journal of Shenyang Institute of Aeronautical Engineering,2009,26(3):78-81. |
| |
| Authors: | GUO Li LIU Lei JI Duo CAI Dong-feng |
| |
| Institution: | Knowledge Engineering center;Shenyang Institute of Aeronautical Engineering;Liaoning Shenyang 110034 |
| |
| Abstract: | Relational similarity algorithm has important application in field of the information retrieval.At present,relational similarity is computed through Vector Space Model mostly,but data sparse impacts the relational similarity calculation result badly.Therefore,a pattern to express the characteristic and singular value decomposition algorithm also is used to reduce the number of dimensions,and the relational similarity is calculated through cosine.The result show that the algorithm is significantly improved c... |
| |
| Keywords: | relational similarity vector space model singular value decomposition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
