【作者】 陶富民； 高军； 周凯；

【Author】 Tao Fumin,Gao Jun,Zhou Kai Network and Information Systems,EECS,Peking University,Beijing,100871

【机构】 北京大学数据库实验室；

【摘要】 情感特征的提取是进行文本的情感分析的一个非常重要的步骤,也是影响其结果好坏的主要因素。在本文中,作者提出一种新的特征提取的方法来解决新闻评论的情感分析问题。在该方法中,首先根据评论和新闻的对比分析获得的候选情感特征,然后经过相关的扩充和验证操作得到通用的情感特征,并将其用于新闻评论的情感分析。对新闻进行话题划分后可以进行更细粒度的情感分析:根据新闻话题信息,设计相应的话题相关的特征对比和验证过程,选取出面向话题的情感特征,最后用面向话题的情感特征对相应话题的进行情感分析。实验证明,这种情感特征提取的方法对于新闻评论这种语句短,评论对象相对分散的评论的情感分析效果有较大的改进。更多还原

【Abstract】 Feature extraction plays a vital role in the area of text based sentiment analysis and is the most important factor in order to obtain high quality analysis.In this paper,the authors proposed a novel approach to extract sentimental features for news comments’ sentiment analysis.Firstly,the candidate sentimental features would be extracted based on the comparison between contents of the news comments and the corresponding news. After that,the general sentimental features,which could be used for sentiment analysis on any general news comment,are selected by applying several extension and validation processes. What’s more,the most significant feature of this proposed method is that it can provide finer-grained sentimental analysis for specific news topic.Specifically,based on the topic information of news comments,it designed corresponding feature comparison and validation policies to extract topical sentiment features and finally utilized these topic specific sentiment features in the sentiment analysis.The experiments have shown that this new proposal would guarantee high performance for the sentiment analysis in sparse data sets,such as comments from news.更多还原