摘要
提出一种基于句子相关度的文本自动分类模型(TCSC).该模型利用训练样本增量式地自动更新类别语料库,根据句子的位置权值和语料权值计算句子类别相关度,获得用于文本分类的句子相关度矩阵,通过该矩阵实现文档分类.该模型避免了分类阶段待分类文本特别是中文文本的分词,模糊了词的多义问题,且在文本分类的实验中能够达到86%以上的查全率和查准率;随着语料库的不断训练和调整,分类性能还可以进一步提高,具有简单实现的特点.
A text category model based on sentence correlation（TCSC） was presented , which incrementally updates category corpus with the training documents automatically. Then, category correlation was obtained by means of sentence position weight and corpus item weight to achieve correlation matrix for text classification. This model avoids the problem of word segmentation in Chinese documents and lowers the effect of words with multiple meanings in the phase of classification. Experimental results show that the recall and precision of this model reached of over 86%, and can be improved by updating corpus. This model can also be implemented easily in programming.
基金
国家自然科学基金(69835010)资助
关键词
文本分类
语料库
相关度矩阵
句权
text-classification
corpus
sentence correlation matrix
sentence weight
