dmis-lab/ASH: Source Code and Data for IEEE TBD Submission: ASH Framework

This release contains the source code and dataset for the manuscript submitted to IEEE Transactions on Big Data. Paper Title: ASH: A Robust Automated Evaluation Framework for Creative Text Generation in Cuisine Transfer Tasks Authors: Hoonick Lee, Mogan Gim, Donghyeon Park, Donghee Choi, Jaewoo Kang Contents Code: Implementation of the ASH evaluation framework, prompt engineering strategies, and meta-evaluation scripts. Data: data/generation: 4,800 LLM-generated recipes across 40 cuisines. data/human: Human ground truth annotations for validation. Note: Full evaluation logs (>800MB) are excluded due to size constraints but can be reproduced using the provided scripts. Usage Please refer to the README.md for installation and execution instructions.