【作者】 黄进；

【导师】 阮彤；

【作者基本信息】 华东理工大学 ， 计算机应用技术， 2014， 硕士

【摘要】 互联网上充斥着大量用户生成数据(User Generated Content, UGC),典型的如论坛。金融领域比较多的是股吧中的评论和研究者的个股点评,股评观点通常有很强的倾向性。研究发现用户的情感倾向与市场的波动有着紧密联系,对金融市场的监控和股价异常处理有着重要作用,因此对金融领域用户生成的文本进行情感分析就显得很有意义。本文根据金融领域文本的特点,从基于语义的方法对web文本进行情感分析,充分考虑到金融领域情感词的特征、句子中词语的位置关系以及相互间的修饰关系,利用SVM结合Stanford句法依存分析,对每个文档计算出一个情感值。面向金融领域文本情感分析,在三个方面进行了创新：1)将情感词在单个句子中的位置关系考虑进去；2)构建了金融领域情感词之间相互修饰的依存分析模板；3)提出了SVM结合依存分析的文本情感倾向性分析算法,并在金融领域中达到较高的查准率和查全率。实验中我们通过从一些重要财经网站上抽取的金融领域数据进行实验,综合值F达到了84.1%,结果表明该方法可以有效的运用于金融领域的文本情感分析。更多还原

【Abstract】 Internet is flooded with a large number of user-generated data (User Generated Content, UGC), such as the typical forum. The financial sector is relatively more in the comments of stocks and researchers Stock Review. Stock analysts view is usually with a strong bias. At the same time, it was found that research users emotional tendencies to monitor and handling the market price of exception plays an important role, so the sentiment analysis of user-generated text in financial sector becomes meaningful.Based on the characteristics of financial sector, we first use the semantic-based approach to web text sentiment analysis, followed by using the SVM with Stanford syntactic dependency analysis which fully into account the character-istics of emotional words, the position and the relationship of mutual modification. For the text sentiment analysis in financial sector, we have innovation in three aspects:1) take positional relationship of the emotional words in the sentence into account;2) constructed dependency parsing template in the financial domain which between emotional words;3) proposed SVM combines dependency parsing analysis algorithms, and achieve high precision and recall in the financial domain.Through the experiment which based on extracted data from a number of important financial sites, the integrated value of F reached84.1%and the results show that this method can be effectively applied in the text sentiment analysis in financial sector.更多还原