This is a brief summary of paper for me to study and arrange for Word-Context Character Embeddings for Chinese Word Segmentation (Zhou et al., EMNLP 2017) I read and studied.

This paper is a research ralted to chinese segmenataion for cross domain.

They also used label embedding to use segmentation label information in the pre-training of character embedding

The model below is the baseline for Chinese Word segmenatation task they used.

Zhou et al., EMNLP 2017

They are inspired by skip-gram embedding model for pre-trainig of word-context character embedding.

Like image below, in order to train word-context character embedding, they used the context of character with the window size c, together with their corresponding segment labels.

Zhou et al., EMNLP 2017

Reference