Ensemble Model for Multi-Source Cross-Domain Sentiment Classification with Little Labeled Data

Kwun Ping Lai, Jackie Chun-Sing Ho, Wai Lam

Abstract: We investigate the problem task of multi-source cross-domain sentiment classification with little labeled data, which is a common problem faced by some online store owners. We propose a knowledge-diverse ensemble model which is capable of automatically capturing important topical knowledge used for cross-domain sentiment classification. Through multiple stages of constructing new pseudo-data for training, it can maintain the diversity of the captured knowledge by different base learners. Experiments on a real-world product review dataset show that our proposed model has a good performance even under the little labeled data constraint.

External IDs:dblp:conf/webi/LaiHL20