A Chinese diplomatic rhetoric dataset (DRD) for supervised fine-tuning of large language models

<p indent="0mm">The dataset is derived from the routine press conferences held by the spokespersons of China′s Ministry of Foreign Affairs between 2000 and 2024. A total of 20,745 Q&amp;A pairs were collected and curated, resulting in a comprehensive Chinese Diplomatic Rhetoric Dataset (DRD) intended for supervised fine-tuning of large language models. The aim is to provide a specialized dataset on diplomatic dialogue strategies to improved existing Chinese language models, enabling them to accurately understand and respond to diplomatic discourse within an international context. The study introduces the N-Jaccard text similarity algorithm, which reduces sensitivity to text length and takes into account the order of words, thereby capturing semantic relationships and logical connections more effectively. Utilizing the GPT-3.5 Turbo large language model, key information was extracted from the data, transformed into contextually appropriate topics, and refined to generate input prompts and output responses. Both machine and human reviewers audited and corrected the initially processed data to ensure the dataset’s quality. The DRD dataset, based on extensive high-quality Q&amp;A data, effectively demonstrates the conciseness, accuracy, and artistry of diplomatic language expression.