用于方面情感分析的结合图卷积神经网络的注意力模型

特定方面情感分析旨在自动识别同一句子中不同方面的情感极性。在现有方法中，结合注意力机制的循环神经网络模型在特定方面情感分析任务中取得了较好的效果，但是大多都忽略了句子的句法特征。因此，该文提出一种结合图卷积神经网络的注意力网络(GCN-aware Attention Networks, GCAN)模型。首先通过长短时记忆网络获取句子的序列信息，并利用图卷积神经网络来捕获语义特征，然后提出两种特征融合方式，得到基于序列信息和语义特征的特定方面表示。在此基础上，引入双向注意力机制处理特定方面包含多个单词的情况，进而获得更精准的基于特定方面的上下文表示。与ASGCN模型相比，该方法在Twitter数据集和SemEval14/15数据集上的分类准确率分别提升了0.34%、0.94%、1.43%和1.23%,F₁值分别提升了0.53%、1.55%、1.60%和2.54%,验证了GCAN的有效性。

GCN-aware Attention Networks for Aspect-based Sentiment Analysis

Aspect-based sentiment analysis is to automatically identify the sentiment polarity to a specific aspect given a context sentence. To fully capture the syntactic information in the sentence, this paper presents a novel GCN-aware Attention Networks (GCAN) for aspect based sentiment analysis. In our method, the aspect-specific representations are generated by combining the sequence information captured by LSTM and syntactic feature captured by the GCN. A bidirectional attention mechanism is tailored for solving the aspect with multi-words and generates the final contextual representation based on specific aspect. Compared with ASGCN model, the classification accuracy is increased by 0.34%,0.94%,1.43% and 1.23%, and the F₁ by 0.53%,1.55%,1.60% and 2.54%, respectively on Twitter dataset and SemEval14/15 dataset.