This repo contains the data of the following paper:

Solving Chinese Character Puzzles Based on Character. Da Ren, Yi Cai, et.al. NLPCC 2019.

This paper proposes a Stroke Sensitive Character Guessing (SSCG) Model which can solve Chinese character puzzles.We collect Chinese character puzzles from Baidu Hanyu and Hydcd. Each character puzzle has a corresponding answer. The strokes of each word is collected from Httpcn.

• The train/valid/test.csv contain the Chinese puzzle-answer pairs. For each row, a puzzle description corresponds to a correct answer.
• As for the stroke.csv, it contains the word-stroke information. We encode the stroke as the number index. Different number represents different stroke.

@InProceedings{10.1007/978-3-030-32233-5_24,
    author="Ren, Da
    and Cai, Yi
    and Li, Weizhao
    and Xia, Ruihang
    and Li, Zilu
    and Li, Qing",
    editor="Tang, Jie
    and Kan, Min-Yen
    and Zhao, Dongyan
    and Li, Sujian
    and Zan, Hongying",
    title="Solving Chinese Character Puzzles Based on Character Strokes",
    booktitle="Natural Language Processing and Chinese Computing",
    year="2019",
    publisher="Springer International Publishing",
    address="Cham",
    pages="303--313",
    isbn="978-3-030-32233-5"
}