A database for automatic recognition of anthropomorphic rhetoric

After filtering out special genres such as poetry, drama, and classical Chinese, 524 articles were used and annotated to form 2329 experimental rhetorical texts, including 1329 anthropomorphic texts and 1000 non anthropomorphic texts. This dataset consists of two parts: the Chinese language textbook data compiled by the department and the corpus for identifying and annotating anthropomorphic figures of speech.

The data of the Chinese language textbook in the Ministry's edition has been classified by grade. The corpus of anthropomorphic speech recognition annotation contains 5 files, including an explanation of how to use the data, a training set for anthropomorphic speech recognition models (used for model training and learning), a validation set for anthropomorphic speech recognition models (used for validation when adjusting model parameters), and a testing set for anthropomorphic speech recognition models (used for model ability testing).