Fine-Grained Classification of Named Entities by Fusing Multi-Features

Due to the increase in the number of classes and the decrease in the semantic differences between classes, fine-grained classification of Named Entities is a more difficult task than classic classification of NEs. Using only simple local context features for this fine-grained task cannot yield a good classification performance. This paper proposes a method exploiting Multi-features for fine-grained classification of Named Entities. In addition to adopting the context features, we introduce three new features into our classification model: the cluster-based features, the entityrelated features and the class-specific features. We experiment on them separately and also fused with prior ones on the subcategorization of person names. Results show that our method achieves a significant improvement for the fine-grained classification task when the new features are fused with others.