// the find
liuhuanyong/ChineseSemanticKB
ChineseSemanticKB,chinese semantic knowledge base, 面向中文处理的12类、百万规模的语义常用词典，包括34万抽象语义库、34万反义语义库、43万同义语义库等，可支持句子扩展、转写、事件抽象与泛化等多种应用场景。
A static Chinese semantic lexicon collection: 12 dictionary types covering synonyms, antonyms, abstractions, abbreviations, stopwords, and modal words, totaling around one million entries. Built by a Chinese Academy of Sciences researcher from public text, manual curation, and automated extraction. Useful as cold-start vocabulary resources for Chinese NLP tasks like query expansion, sentiment analysis, and event abstraction.
The abstraction hierarchy (抽象关系库) is the genuinely interesting piece — 346k entries mapping specific terms to hypernyms, which is hard to find openly for Chinese. Scale is real: 424k synonym pairs and 136k abbreviation mappings are large enough to be practically useful. All data is plain text files you can load without any tooling or database setup. The author has a track record of maintaining Chinese NLP resources, so this isn't a one-off dump.
Last updated March 2023 and effectively unmaintained — no issue responses, no version history for the dict files, no changelog. No quality metrics or evaluation methodology documented; the README admits quality is inconsistent and asks users to report errors, which is not reassuring for production use. The Python repo contains almost no code — just an `__init__.py` and the text files — so there's no built-in lookup API, deduplication, or tooling. It also covers modern slang and internet Chinese poorly, which limits usefulness for social media or contemporary text analysis.