项目名称: 文本情绪分析中的关键问题研究
项目编号: No.61203378
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 徐军
作者单位: 哈尔滨工业大学
项目金额: 25万元
中文摘要: 由于受到智能人机交互、舆情分析和产品推荐系统等新型应用需求驱动,文本的情绪分析研究日显重要。本项目从不同角度出发,围绕文本情绪分类、读者情绪归因和读者情绪预测三个问题进行深入研究。首先从文本作者和文本描述对象角度,研究结合机器学习技术与情绪表达模式的方法,分析文本中包含、表达的情绪状态;其次,基于大量新闻文本的读者情绪社会化标注语料,研究利用统计、主题模型和聚类等手段,以词语反射和事件主题为线索,对读者情绪反馈进行自动归因研究,进而,结合自动归因研究,研究集成分类器和多标签主题模型对新闻文本的读者情绪反馈进行预测的方法。本项目的实施为探索文本情绪表达、理解和预测及其在自然语言处理任务中的应用奠定基础。
中文关键词: 情绪分析;情绪分类;情绪归因;情绪预测;情感分析
英文摘要: With the rapid development of intelligent human-machine interaction, public opinion monitoring and online product recommendation, text emotion analysis has attracted growing attentions. From different perspectives, this project studies three key problems in Chinese emotion analysis i.e. emotion classification of text, reader's emotion cause detection and reader's emotion prediction. Firstly, the method incorporating machine learning technology and emotion expression patterns is investigated to classify text emotions. Secondly, based on a large scale of reader emotion feedback corpus, the approach which incorporates lexical statistics, topic model and cluster technology is investigated to discover the causes of reader's emotion while the word and event reflection are used as clues. Finally, a multi-view ensemble classifier and a multi-labeled LDA are proposed for reader's emotion prediction based on the cause detection research. This project explores the text emotion expression, understanding and prediction text systematically. It is a foundation to emotion-related natural language processing applications.
英文关键词: Emotion Analysis;Emotion Classification;Emotion Reasoning;Emotion Prediction;Sentiment Analysis