基于压缩稀疏矩阵矢量相乘的文本相似度计算

Document Similarity Degree Measuring Based on Compressed Sparse Matrix Vector Multiplication Technique

在信息检索矢量模型的基础上,提出了一种基于压缩稀疏矩阵矢量相乘的文本相似度计算方法,具有矢量模型计算简单和速度快的特点.该方法采用压缩稀疏矩阵矢量空间存储数据,在相似度计算和数据存储时不需要考虑文本矢量矩阵中的零元素,大大减少了计算量和存储空间,从而使信息检索系统运行效率显著提高.仿真实验表明,上述方法比基于矢量模型的传统反向索引机制节省了38%的存储空间.

A novel method to measure document similarity degree based on compressed sparse matrix vector multiplication technique was presented. The designe of the method is based on an information retrieval vector model, which has the virtues of beeing simple and high speed. A compressed sparse matrix vector space is used, in which the zero elements in the vector matrix of documents are not processed while calculating similarity degree and storing data, to reduce the requirements of calcualting time and storing space. This method can improve the efficiency of information retrieval system. Simulation experiment indicats that the method can save 38% of the storing space of the conventional inverted index technique based on vector model.