HowNet

该数据集名为HowNet，是一个中文语义知识库，其中单词与它们的语义素进行标注，语义素是人类语言中概念的最小语义单元。该数据集还包括语义素的分层结构，用于提高自然语言处理任务中的标注效率。规模上，该数据集包含1400个语义素，以及103,843个单词，这些单词被标注了212,539个义项。其应用的任务主要是语义素预测。

This dataset, named HowNet, is a Chinese semantic knowledge base where words are annotated with their corresponding semantic primitives—semantic primitives refer to the minimal semantic units of concepts in human language. The dataset also includes a hierarchical structure of semantic primitives, which is designed to improve annotation efficiency in natural language processing (NLP) tasks. In terms of scale, this dataset contains 1,400 semantic primitives, 103,843 words, and these words are annotated with 212,539 word senses. The primary application task of this dataset is semantic primitive prediction.