【摘要】 社交文本的情感分析主要存在结构不规则、特征稀疏和分类效果不理想等问题。针对这些问题，提出了一种双向长短期记忆网络(Bi-directional long short-term memory, BiLSTM)和胶囊网络(Capsule network, CapsNet)混合模型(BiLSTM-CapsNet)。该模型先使用胶囊网络提取单个特征词在整个句子中的位置语义信息，再使用双向长短期记忆网络提取社交文本的上下文词语之间的关系，最后通过softmax分类器，进行情感倾向的分类。试验结果表明，该模型在粗粒度和细粒度情感分类中都有优势。更多还原

【Abstract】 The sentiment analysis of social text mainly has problems such as irregular structure, sparse features and unsatisfactory classification effect. In response to these problems, a hybrid model of bi-directional long short-term memory network and a capsule network(BiLSTM-CapsNet)is proposed. The model first uses the capsule network(CapsNet)to extract the position semantic information of a single feature word in the whole sentence, and then uses the bi-directional long short-term memory(BiLSTM)to extract the relationship between the contextual words of social text and finally uses the softmax classifier to classify the sentiment tendency. Experimental results show that the model has advantages in both coarse-grained and fine-grained sentiment classification.更多还原