Zhao, Y.; Wu, H.; Xu, B. Leveraging Attention to Effectively Compress Prompts for Long-Context LLMs. AAAI 2025, 39, 26048-26056.