资料文稿案例大数据8a9504a.pdfVIP

2015 IEEE International Conference on Data Mining2015 IEEE International Conference on Data Mining Collaborative Multi-Domain Sentiment Classification zhao Wu, Yongfeng Huang Tsinghua National Laboratory for Information Science and Technology, Department of Electronic Engineering, Tsinghua University, , : wu zhao@, yfhuang@ Abstrac timent classification is a hot research topic in content is also proven to be useful for user interest mining, both industrial and academic fields. The mainstream sentiment alized recommendation, social advertising, customer re- classification methods are based on machine learning and treat lation management and crisis management [29], [32]. Thus sentiment classification as a text classification problem. However, sentiment classification has been a hot topic in both industrial sentiment classification is widely recognized as a highly domain- and academic fields [26], [17], [18], [31]. dependent task. The sentiment classifier trained in one domain may not perform well in another domain. A simple solution to The mainstream method of sentiment classification is this problem is training a domain-specific sentiment classifier for regarding it as a text classification problem and applying each domain. However, it is difficult to label enough data for