项目名称: 短文本情感分析关键技术研究
项目编号: No.61502478
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 计算机科学学科
项目作者: 林政
作者单位: 中国科学院信息工程研究所
项目金额: 21万元
中文摘要: 情感分析具有广阔的应用前景,可以带来巨大的经济和社会效益。随着社交网络的蓬勃发展,短文本具有数据量大、内容简略、特征稀疏、信息混杂等特点,这使得以往的情感分析方法在处理短文本时,难以保证其分析效果。针对上述挑战,本课题按照短文本情感要素抽取、短文本情感分类、短文本情感归纳三层不同的研究任务,构造一系列情感分析模型,分别解决隐式属性词及基于属性词的情感知识获取问题、情感词典抽取过于依赖外部资源的问题、短文本情感分类的数据稀疏问题、短文本情感归纳的表征与建模问题。本项目所研究的这一系列模型能够自底向上满足不同层次情感分析的实际需求。
中文关键词: 情感分析;观点挖掘;情感分类;情感词
英文摘要: There are broad prospects in sentiment analysis field which can bring huge economic and social effects. With the development of social network, short texts are characterized by the features like big data volume, concise content, sparse characteristics, mixed information and etc., which make previous sentiment analysis approaches difficult to achieve better results. To tackle the above challenge, we construct a series of sentiment analysis models in the manner of sentiment elements extraction, sentiment classification and sentiment summarization for short text. The problems that can be solved in this project include: implicit aspect discovery and aspect-dependent sentiment knowledge extraction; over reliance on external resources in opinion lexicon extraction; data sparsity in sentiment classification for short text; representation and modeling of sentiment summarization for short text. The models proposed in this project can satisfy all needs of sentiment analysis from low level to high.
英文关键词: sentiment analysis;opinion mining;sentiment classification;opinion word