摘 要:为了解决如何获取文本的位置信息和捕获文本中更多情感信息的问题,利用一种结合 ALBERT 模型和卷积神经网络 CNN 的外卖评论情感分析模型 ALBERT-CNN 的方法。首先,ALBERT 对文本词向量表示,获得文本动态特征表示;其次,利用卷积神经网络 CNN 对特征进行训练,有效获取更丰富的局部信息;最后,对 ALBERT 和 CNN 进行融合后提取的特征通过 Softmax 函数对外卖评论文本进行情感分类,并使用 R_Drop 对模型进行正则化。实验结果表明,与传统模型相比,使用了 R_Drop 的 ALBERT-CNN 模型的精确度 P、召回率 R 和 F1 值均有提高。
关键词:外卖评价;评论文本;情感分析;ALBERT-CNN 模型;R_Drop
DOI:10.19850/j.cnki.2096-4706.2022.10.040
基金项目:安徽理工大学 2020 年研究生创新基金项目(2020CX2071)
中图分类号:TP312 文献标识码:A 文章编号:2096-4706(2022)10-0157-04
Emotional Analysis of Takeout Comments Based on ALBERT-CNN
HU Shengli, ZHANG Liping
(School of Computer Science and Engineering, Anhui University of Technology and Technology, Huainan 232001, China)
Abstract: In order to solve the problem of how to obtain the location information of the text and capture more emotional information in the text, a method of takeout comment emotional analysis model ALBERT-CNN combining ALBERT model with convolutional neural network CNN is used. Firstly, ALBERT represents the text word vector to obtain the text dynamic feature representation. Secondly, the convolutional neural network CNN is used to train the features to effectively obtain richer local information. Finally, the features extracted after the fusion of ALBERT and CNN are emotionally classified by Softmax function, and R_Drop is used to regularize the model. The experimental results show that compared with the traditional model, the accuracy P, recall R and F1 values of the ALBERT-CNN model using R_Drop are improved.
Keywords: takeout evaluation; comment text; emotion analysis; ALBERT-CNN model; R_Drop
参考文献:
[1] LIANG H Z,GANESHBABU U,THORNE T.A dynamic ba yesian network approach for analysing topicsentiment evolution [J].IEEE Access,2020,8:54164-54174.
[2] 王美荣 . 基于卷积神经网络的文本分类算法 [J]. 佳木斯大学学报(自然科学版),2018,36(3):354-357.
[3] ZHANG X J,HUANG S,ZHAO J Q,e t al.Exploring deep recurrent convolution neural networks for subjectivity classification [J].IEEE Access,2019,7:347-357.
[4] PAN Y,LIANG M.Chinese Text Sentiment Analysis Based on BI-GRU and Self-attention [C]//2020 IEEE 4th Information Technology,Networking,Electronic and Automation Control Conference (ITNEC). Chongqing: IEEE,2020:1983-1988.
[5] 赵富,杨洋,蒋瑞,等 . 融合词性的双注意力 Bi-LSTM情感分析 [J]. 计算机应用,2018,38(S2):103-106+147.
[6] DEVLIN J,CHANG M W,LEE K,et al.BERT:Pre-training of Deep Bidirectional Transformers for Language Understanding [J/OL].arXiv:1810.04805 [cs.CL] .[2022-03-04].https://arxiv.org/abs/1810.04805v1.
作者简介:胡胜利(1978—),男,汉族,安徽淮南人,副教授,硕士,研究方向:信息智能计算。