EhiMeNLP at TSAR 2025 Shared Task Candidate Generation via Iterative Simplification and Reranking by Readability and Semantic Similarity

AbstractWe introduce the EhiMeNLP submission, which won the TSAR 2025 Shared Task on Readability-Controlled Text Simplification. Our system employed a two-step strategy of candidate generation and reranking. For candidate generation, we simplified the given text into more readable versions by combining multiple large language models with prompts. Then, for reranking, we selected the best candidate by readability-based filtering and ranking based on semantic similarity to the original text.