Page 174 - 《软件学报》2021年第9期

2798 Journal of Software 软件学报 Vol.32, No.9, September 2021

5 总结和未来工作

本文提出了一个具有自注意力机制和多通道特征的双向 LSTM 模型(MFSA-BiLSTM).该模型由自注意力
机制和多通道特征两部分组成.先对情感分析任务中现有的语言知识和情感资源进行建模,生成不同的特征通
道作为模型的输入,再利用 BiLSTM 来充分的获得这些有效的情感资源信息.最后使用自注意力机制对这些重
要信息进行重点关注加强,提高分类精度.此外,本文在 MFSA-BiLSTM 模型上,针对文档级文本分类任务提出了
MFSA-BiLSTM-D 模型.该模型将文本中的句子进行分割,再分别使用 MFSA-BiLSTM 模型进行特征学习得到
句子特征信息.在 5 个基准数据集上进行了实验,用来评估本文提出的方法的性能.实验结果表明:在大多数情况
下,MFSA-BILSTM 和 MFSA-BILSTM-D 模型比一些最先进的基线方法分类更好.
未来的工作重点是注意力机制的研究和文档级文本特定目标分类任务的网络模型体系结构的设计.未来
的工作主要包括以下几个部分:(1) 利用其他注意机制进一步完善本文提出的方法;(2) 针对文档级文本特定目
标分类任务,设计了一种新的注意机制和网络模型;(3) 将本文的方法应用到实际应用中.

References:
[1] Socher R, Pennington J, Huang EH, Ng AY, Manning CD. Semi-supervised recursive autoencoders for predicting sentiment
distributions. In: Proc. of the 2011 Conf. on Empirical Methods in Natural Language Processing. Stroudsburg: Association for
Computational Linguistics, 2011. 151−161.
[2] Socher R, Perelygin A, Wu J, Chuang J, Manning CD, Ng A, Potts C. Recursive deep models for semantic compositionality over a
sentiment treebank. In: Proc. of the 2013 Conf. on Empirical Methods in Natural Language Processing. Stroudsburg: Association
for Computational Linguistics, 2013. 1631−1642.
[3] Kim Y. Convolutional neural networks for sentence classification. In: Proc. of the 2014 Conf. on Empirical Methods in Natural
Language Processing (EMNLP). Stroudsburg: Association for Computational Linguistics, 2014. 1746−1751. [doi: 10.3115/v1/D14-
1181]
[4] Kalchbrenner N, Grefenstette E, Blunsom P. A convolutional neural network for modelling sentences. In: Proc. of the 52nd Annual
Meeting of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2014. 655−665.
[doi: 10.3115/v1/P14-1062]
[5] Lei T, Barzilay R, Jaakkola T. Molding CNNs for text: Non-linear, non-consecutive convolutions. In: Proc. of the 2015 Conf. on
Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2015. 1565−1575.
[doi: 10.18653/v1/D15-1180]
[6] Zhu X, Sobihani P, Guo H. Long short-term memory over recursive structures. In: Proc. of Int’l Conf. on Machine Learning. 2015.
1604−1612.
[7] Tai KS, Socher R, Manning CD. Improved semantic representations from tree-structured long short-term memory networks. In:
Proc. of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th Int’l Joint Conf. on Natural
Language Processing. Stroudsburg: Association for Computational Linguistics, 2015. 1556−1566. [doi: 10.3115/v1/P15-1150]
[8] Chen Z, Xu RF, Gui L, Lu Q. Combining convolutional neural networks and word sentiment sequence features for Chinese text
sentiment analysis. Journal of Chinese Information Processing, 2015,29(6):172−178 (in Chinese with English abstract). http://jcip.
cipsc.org.cn/CN/Y2015/V29/I6/172 [doi: CNKI:SUN:MESS.0.2015-06-024]
[9] Qian Q, Huang M, Lei J, Zhu X. Linguistically regularized LSTM for sentiment classification. In: Proc. of the 55th Annual Meeting
of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2017. 1679−1689. [doi:
10.18653/v1/P17-1154]
[10] Pei SW, Wang LL. Text sentiment analysis based on attention mechanism. Computer Engineering and Scienc, 2019,41(2): 344−353
(in Chinese with English abstract). [doi: CNKI:SUN:JSJK.0.2019-02-023]
[11] Liu G, Guo J. Bidirectional LSTM with attention mechanism and convolutional layer for text classification. Neurocomputing, 2019,
337:325−338. [doi: 10.1016/j.neucom.2019.01.078]
[12] Tang D, Wei F, Yang N, Zhou M, Liu T, Qin B. Learning sentiment-specific word embedding for twitter sentiment classification.
In: Proc. of the 52nd Annual Meeting of the Association for Computational Linguistics. Stroudsburg: Association for
Computational Linguistics, 2014. 1555−1565. [doi: 10.3115/v1/P14-1146]