Abstract: This paper proposes a sentence space model(SSM)and a classification algorithm based on SSM. It compares the vector space model and the sentence space model in classifying the same document with the recall and the precision from different granularity, word granularity and sentence granularity. Experiments show SSM has a better classification performance than vector space model in many circumstances.

Key words: Granularity, Vector space model, Sentence space model

摘要： 提出了句子空间模型及基于句子空间模型的分类算法。比较了从词、句子两个不同粒度对文档进行表示的向量空间模型和句子空间模型在对同一问题进行分类时的召回率和准确率。实验表明，与向量空间模型相比，句子空间模型在许多情况下具有较好的分类性能。

关键词: 粒度, 向量空间模型, 句子空间模型