Abstract: Large language models (LLMs) have made significant advancements in natural language processing, yet they still face challenges in tasks requiring deep comprehension of structurally complex ...