CCNC

Dataset Overview

Basic Statistics

Pinyin Version

• Surname pinyin uses a custom Chinese Surname Phonetic Dictionary; given names use pypinyin.
• Three versions of the corpus are available for download: pure Chinese version and two pinyin versions.

Sources

• Dataset origins:
  • Name Encyclopedia: contributes 2,513,097 entries.
  • Chinese Personal Names Corpus: contributes 1,145,012 entries.
• Processing details:
  • Distinguished names that were originally unsegmented.
  • Removed approximately three hundred thousand duplicate entries.
  • Treated same-name different-gender instances as separate entries.
  • Unknown-gender entries are from the Chinese Personal Names Corpus.

Chinese Surname Phonetic Dictionary

• Contains 1,606 Chinese surnames with their pinyin.
• 1,534 surnames and their phonetics are sourced from Mingba Baijia Xing; the remaining 72 were manually annotated by the author.

Train/Test/Predict Splits

• Provides code to split the corpus into training, testing, and prediction sets, default ratio 6:2:2.
• Pre-split full Chinese character version compressed file download link: full Chinese character version.