融合反讽语言特征的反讽语句识别模型

Ironic sentence recognition model integrating ironic language features

反讽是采用内隐的形式来表达情感的一种方法,反讽语句在文字和所想表达的情感上存在着不同,这使得对反讽语句进行情感分类变得更加困难。针对这一现象,提出一种融合反讽语言特征的反讽语句识别模型,通过加入反讽语言特征来提高反讽语句的识别准确率。首先,采用卡方检验算法对反讽语言进行分析并获取语言特征;然后,利用Word2Vec对语言特征进行训练获取语言特征的特征表示,同时使用注意力机制与Bi-GRU(双向门控循环神经单元)模型获取句子的特征表示;最后,将语言特征的特征表示与句子的特征表示进行融合并作为情感分类层的输入,对反讽语句进行识别。与CNN-AT、CNN-Adv、EPSN等3种模型进行对比,实验结果表明,该模型可以有效提高对于反讽语句的识别准确率。

Irony is a method of expressing sentiment implicitly.Differences between the words and the emotions of ironic sentences are abundant,causing difficulty in the sentiment classification of ironic sentences.To solve this problem,an ironic sentence recognition model integrating ironic language features(ISR)is proposed to improve the recognition accuracy of the ironic sentence by adding ironic language features.Initially,the Chi-square test algorithm is used to analyze ironic language and obtain language features.Then,Word2Vec is used to train the language features to obtain the feature representation of the language features.At the same time,the attention mechanism and Bi-GRU(bidirectional gated recursive neural unit)model are used to obtain the feature representation of the sentence.Finally,the feature representations of language features and sentences are fused as the input of the sentiment classification layer to identify the ironic sentences.The model has been compared with CNN-AT,CNN-Adv,and EPSN models.Experiment results show that the proposed model has high recognition accuracy for the ironic sentence.