中文摘要：

针对传统的文本分类方法需要大量人工标注好的训练数据,且数据标注的好坏会影响结果等问题,通过对LDA及其相关模型的研究,提出一种基于LDA的弱监督文本分类算法。无需人工标注训练数据,在处理文本时,引入词向量,保持文本中的词序,加入二元语法。实验结果表明,该方法节省了人力、物力,取得了较优效果。

英文摘要：

To resolve problems that the traditional text classification methods need a lot of manually labeled training data and that the quality of the data influences the results, through the study of LDA and its related models, a weakly supervised text classifi- cation algorithm on the basis of LDA was presented. Manually labeled training data were no longer needed. Besides, when dealing with the text, the word vector was introduced, and the word order was maintained and the bigram grammar was joined. Experimental results show that when this approach reduces manpower and material resources, it also obtains better effects.