Fine-tuned RoBERTa Model Weights and Code for Sentiment Analysis of "Journey to the West" English Translations.

Abstract: This archive provides a comprehensive computational package for the research on the cross-cultural reception and sentiment analysis of Journey to the West English translations. It contains the complete end-to-end workflow, including raw/processed datasets, fine-tuned model weights (Group 2, SST-5), and the full suite of scripts for data processing and advanced visualization. This repository ensures full transparency and reproducibility for the manuscript submitted to PLOS ONE. Contents of the ZIP Archive: Full Datasets: Includes all raw reader review data, pre-processed sentence-level datasets, and final sentiment-labeled outputs used in the study. Fine-tuned Model (Group 2): Complete weights (model.safetensors) and configuration files based on the RoBERTa architecture, specifically optimized on the SST-5 dataset for multi-class sentiment classification. Source Code & Visualization Scripts: A comprehensive collection of Python scripts covering data cleaning, model training, batch inference, and the generation of high-quality statistical visualizations (e.g., sentiment distributions, thematic maps, and emotional landscapes). Visual Outputs: Generated figures and charts depicting the reception trends and emotional nuances of different English translations.