With the rapid growth of the big data, many approaches in the representation of text for sentiment classification have been successfully proposed in natural language processing. However, these approaches remedy this problem based on single-task supervised objectives learning and do not consider their relative of multiple tasks. Based on these defects, in this work, we consider these tasks are relative, and use weight-shared parameters for learning the representation of text in neural network model, we introduce and study a multi-task approach with variational auto-encoder generative model (MTVAE) by jointly learning them. Experimental results on six subsets of Amazon review data show that the proposed approach can effectively improve the sentiment classification accuracy by other relative tasks.

随着大数据的快速增长，在自然语言处理中已经成功地提出了许多用于情感分类的文本表示方法。但是，这些方法基于单任务监督目标学习来解决此问题，而不考虑它们与多个任务的相对关系。基于这些缺陷，在这项工作中，我们认为这些任务是相对的，并使用权重共享参数来学习神经网络模型中文本的表示形式，我们引入并研究了一种具有变分自动编码器生成模型的多任务方法（MTVAE）通过共同学习它们。对亚马逊评论数据的六个子集进行的实验结果表明，该方法可以通过其他相关任务有效地提高情感分类的准确性。