Chinese-English Translation Dataset for Humanities and Social Sciences

The dataset comes from Exploring the Construction of Chinese-English Terminology Knowledge Base in the Humanities and Social Sciences: Theory and Methods, published by Nanjing University Press, which is written to meet the needs of two-way information exchange and research in the field of humanities and social sciences in Chinese and English. In order to construct a high-quality dataset to improve the performance of the large language model and to cover more disciplinary categories in the humanities and social sciences, the methods of data expansion and feature extraction are adopted to preprocess the collected data, and to obtain a batch of high-quality Chinese-English terminology cross-referenced data corpus of different disciplinary categories in the field of humanities and social sciences. The collected Chinese-English data are combined to construct a Chinese-English bi-directional dataset. For the experimental work of instruction fine-tuning, a variety of instruction prompts are designed, and different prompts will significantly affect the output results of the model, and the final choice is to incorporate the disciplines of the humanities and social sciences into the structural attribute of instruction as the command fine-tuning. The final choice is to incorporate the disciplines of humanities and social sciences into the structural attribute "instruction" as the prompt for instruction fine-tuning.