Microblogging emotion classification based on multi-classifier integration strategy

This paper proposed a two-step classification approach based on multi-classifier integration strategy for Chinese microblogging emotion classification. The first step identifies the non-emotional microblogging by combining Gradient Boosting Decision Tree (GBDT) and Support Vector Machine (SVM) classifiers. The second step classifies the emotional microblogging into different emotion categories by using the multi-classifier integration approach which combines Multi-label k-Nearest Neighbor (MLKNN+), Support Vector Machine (SVM) and Naive Bayes (NB) classifiers. The evaluations on the NLP&CC013 microblogging analysis dataset shows that our proposed approach obtained promising performance. It achieves the highest performance on this dataset, based on our knowledge.