摘要： 针对文本分类领域中向量空间模型维数过高和空间扭曲的问题，提出一种基于正交分解的新模型。借用物理学中力的正交分解，将高维的文本向量映射到低维的以类别为坐标轴的空间中，解决了高维的向量和扭曲的空间这2个问题。实验表明，与向量空间模型相比，新模型下分类速度有较大提高，精度也有所增加。

关键词: 文本分类, 正交分解, 向量空间模型

Abstract: In text classification area, Vector Space Model(VSM) is the most widely used model while it has two drawbacks: high dimensions and warped space. This paper presents a new model based on orthogonal decomposition. In this model, higher dimensional vectors of texts are mapped in a lower dimensional space which uses categories as its coordinate axes to solve these two drawbacks. Experiment shows that under the new model, the classification process is speeded up to a considerable degree and the precision is increased.

Key words: text classification, orthogonal decomposition, Vector Space Model(VSM)

中图分类号: 

• TP391