A two-stage framework for cross-domain sentiment classification

被引:43
作者
Wu, Qiong [1 ]
Tan, Songbo [1 ]
机构
[1] Chinese Acad Sci, Inst Comp Technol, Key Lab Network, Beijing 100190, Peoples R China
关键词
Sentiment analysis; Opinion mining; Information retrieval; Data mining;
D O I
10.1016/j.eswa.2011.04.240
中图分类号
TP18 [人工智能理论];
学科分类号
140502 [人工智能];
摘要
Supervised sentiment classification systems are typically domain-specific, and the performance decreases sharply when transferred from one domain to another domain. Building these systems involves annotating a large amount of data for every domain, which needs much human labor. So, a reasonable way is to utilize labeled data in one existed (or called source) domain for sentiment classification in target domain. To address this problem, we propose a two-stage framework for cross-domain sentiment classification. At the "building a bridge" stage, we build a bridge between the source domain and the target domain to get some most confidently labeled documents in the target domain; at the "following the structure" stage, we exploit the intrinsic structure, revealed by these most confidently labeled documents, to label the target-domain data. The experimental results indicate that the proposed approach could improve the performance of cross-domain sentiment classification dramatically. (C) 2011 Elsevier Ltd. All rights reserved.
引用
收藏
页码:14269 / 14275
页数:7
相关论文
共 32 条
[1]
ANDO RK, 2005, P ACL
[2]
ANDREEVSKAIA A, 2008, P ACL
[3]
[Anonymous], 2002, P EMNLP
[4]
[Anonymous], P RANLP
[5]
Blitzer J., 2007, P ACL
[6]
Blum A., 1998, Proceedings of the Eleventh Annual Conference on Computational Learning Theory, P92, DOI 10.1145/279943.279962
[7]
Learning from labeled and unlabeled data: An empirical study across techniques and domains [J].
Chawla, N ;
Karakoulas, G .
JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH, 2005, 23 :331-366
[8]
Chelba Ciprian, 2004, P EMNLP
[9]
Cui H., 2006, P AAAI
[10]
DASGUPTA S, 2009, P ACL