This is the dataset adopted in the paper:
	Detection of Chinese Word Usage Errors for Non-Native Chinese Learners with Bidirectional LSTM
	Yow-Ting Shiue, Hen-Hsen Huang and Hsin-Hsi Chen
	Annual Meeting of the Association for Computational Linguistics (ACL 2017)

The file "w_seq_label.txt" contains 10,510 Chinese sentence segments. In each segment, the token enclosed by <> is the ground-truth error position annotated in the HSK corpus. For example, in 
	店 是 爸爸 <留在> 我们 的
, "留在" is the incorrect token. Every segment contains exactly one error position.
