基于迁移学习的微博情绪分类研究
时间:2019-01-12 17:04 来源:毕业论文 作者:毕业论文 点击:次
| 摘要微博允许用户随时随地发表自己的看法和分享所见到的新鲜事，这些新鲜事包含丰富的用户情绪。对微博用户情绪进行分类，能让企业准确把握用户对产品的情感态度，还能帮助政府实时监控民众情绪，文护社会稳定。本文选取了生活领域和H7N9爆发时期两个不同领域的微博语料，用已有的微博标注数据对H7N9爆发时期未标注的微博数据进行情感分类。通过分词、特征选择、特征权重计算、迁移学习NB、IBK、SOM模型等技术，进行主客观分类、情感分类、情绪分类，并检测其准确率。得出了H7N9爆发时期的公众舆论情绪分布，负面情绪多于正面情绪；负面情绪多集中在“愤怒”与“惊讶”方面，但正面情绪“高兴”也因H7N9疫苗的研发成功占很大一部分。得出结论，在公众卫生安全事件爆发的时候，应及时发布相关研究进展对稳定公众情绪。最后针对研究过程中的缺点和不足，本文进行了总结和展望。32623 关键词 情感分析 情绪分类 迁移学习 微博情绪 毕业论文外文摘要 Title Emotion Classification of Micro-blogging Based on Transfer Learning Abstract Weibo allows users to publish their own views and share the new things which contain a wealth of user emotions anytime and anywhere. The classification of the micro blog user emotions can let the enterprise accurately grasp the emotional attitude of the user to the product, but also help the government monitor the public sentiment in time, and maintain social stability. This paper selects the microblog corpus of two different domains in the field of life and the outbreak period of H7N9. The emotional classification of the unlabeled microblog data in the H7N9 outbreak period is carried out with the existing microblog data. Through word segmentation, feature selection, feature weight calculation, migration learning NB, IBK, SOM model and other technologies, subjective and objective classification, emotion classification and sentiment classification are conducted and its accuracy is detected at the same time. The distribution of public opinion in the H7N9 outbreak period is obtained which summmaries negative emotions more than positive emotion. Negative emotions are concentrated in the "anger" and "surprise", but positive emotions "happy" for vaccine successful development are accounted for a large part . It is concluded that, when public health security incident is broken out of, the relevant research should be released to the public in time to stabilize the public mood. At last, this paper summarizes and prospects the shortcomings of the research. Keywords Emotion Analysis, Emotion Classification, Transfer learning, Micro-blog emotion 目录 1 绪论 1 1.1 研究背景 1 1.2 研究意义 2 1.3 研究思路 2 1.4 组织结构 3 2 相关研究综述 5 2.1 情感分析综述 5 2.1.1 不同情感级别的情感分析 5 2.1.2 情感分析研究流程 6 2.2 微博情绪分类概述 7 2.2.1 微博情绪的特点 7 2.2.2 微博情绪分类方法 7 2.2.3 微博情绪分类研究现状 8 2.3 迁移学习在情感分类中的应用 8 2.3.1 迁移学习定义 8 (责任编辑：qin) |