A method of constructing a fine-grained sentiment lexicon for the humanities computing of classical chinese poetry
Wei Zhang,Hao Wang,Min Song,Sanhong Deng
DOI: https://doi.org/10.1007/s00521-022-07690-8
2022-08-24
Neural Computing and Applications
Abstract:In Chinese classics, the sentiment attitudes or thoughts of the ancients regarding specific environments, people, and events were generally expressed in the form of poetry. Compared with previous attempts to classify the polarity of poetry, sentiment terms can be used to detect more fine-grained humanity knowledge in literary information resources. However, the existing techniques of domain sentiment lexicon construction fail to take full advantage of deep learning and linguistic knowledge, which cannot ensure the term integrity and accuracy. To this end, this work proposes a novel approach for the construction of a sentiment lexicon via the combination of supervised sentiment term extraction and classification, aiming at incorporating multi-dimensional linguistic knowledge into a two-phase deep learning model. A character-sequence labeling model for term extraction is first constructed by fusing the emotion radical features of Chinese characters, and term embedding augmentation via word knowledge is then carried out to classify the extracted terms. Experiments on Chinese poetry and its appreciation texts validate the superiority of the proposed method, and the model incorporating linguistic knowledge is found to outperform the benchmark models in different metrics. A fine-grained sentiment lexicon with two first classes, five-second classes, 15 third classes, and 14,368 domain terms and unregistered terms is constructed via hierarchical term classification, thereby contributing to the advancement of the interpretability of the humanities computing of classical Chinese poetry.
computer science, artificial intelligence