Journal of Frontiers of Computer Science and Technology ›› 2021, Vol. 15 ›› Issue (5): 794-811.DOI: 10.3778/j.issn.1673-9418.2010002

• Surveys and Frontiers • Previous Articles     Next Articles

Review of Deep Learning for Short Text Sentiment Tendency Analysis

TANG Lingyan, XIONG Congcong, WANG Yuan, ZHOU Yubo, ZHAO Zijian   

  1. 1. College of Artificial Intelligence,Tianjin University of Science and Technology, Tianjin 300457, China
    2. Population and Precision Health Care, Ltd., Tianjin 300000, China
  • Online:2021-05-01 Published:2021-04-30

基于深度学习的短文本情感倾向分析综述

汤凌燕熊聪聪王嫄周宇博赵子健   

  1. 1.天津科技大学 人工智能学院,天津 300457
    2.普迈康(天津)精准医疗科技有限公司,天津 300000

Abstract:

Short text sentiment tendency analysis is one of the key research issues in the field of natural language processing. Sentiment tendency analysis is a series of methods, techniques and tools used to detect the semantics of subjective inclination contained in language, and it is the key to the deep semantic understanding of text. The randomness, high ambiguity and brevity of short text data make traditional sentiment tendency analysis tasks based on feature engineering and machine learning classification technology limited. With the wide application of deep learning technology in natural language processing, the short text sentiment tendency analysis model based on deep learning has made new breakthroughs. Through combing the relevant literature, this paper first summarizes and compares traditional methods and deep learning methods, introduces and analyzes the short text sentiment  tendency analysis models based on deep learning in recent years, and elaborates the connections, differences and advantages of the models. Second, it summarizes the research hotspots and progress ideas of deep learning in short text sentiment tendency analysis, and the commonly used public datasets and evaluation indicators for sentiment tendency analysis are introduced. Finally, based on the characteristics of deep learning technology and the task difficulties, the application prospect of deep learning in the direction of short text sentiment tendency analysis is predicted.

Key words: sentiment tendency analysis, short text, deep learning

摘要:

短文本情感倾向分析是自然语言处理领域的关键研究问题之一。情感倾向分析是用于检测语言所蕴含主观倾向语义的一系列方法、技术和工具,是对文本深层语义理解的关键。短文本数据的随意性、高歧义性以及简短性使得传统基于特征工程和机器学习分类技术的情感倾向分析任务性能有限。随着深度学习技术在自然语言处理中的广泛应用,基于深度学习的短文本情感倾向分析模型取得了新的突破。通过对相关文献的梳理,首先概述和对比了传统方法和深度学习方法,介绍和剖析了近年基于深度学习的短文本情感倾向分析模型,并阐述了模型的联系、区别与优势;其次归纳了深度学习在短文本情感倾向分析中的研究热点和进展思路,介绍了情感倾向分析常用的公开数据集以及评价指标;最后结合深度学习技术特点和任务难点,对深度学习在短文本情感倾向分析方向的应用前景进行预测。

关键词: 情感倾向分析, 短文本, 深度学习