W-TextCNN: A TextCNN model with weighted word embeddings for Chinese address pattern classification

Chen Zhang, Renzhong Guo, Xiangyuan Ma, Xi Kuai, Biao He

Abstract: Highlights•A fine-grained analysis of address structure is implemented to establish an address pattern classification system.•A novel weighted word embeddings schema is proposed for representing an address by integrating TFIDF and Skip-gram.•A deep learning model W-TextCNN is constructed for address pattern classification.•Address pattern classification contributes to improving segmentation precision and address quality.