KTRL+F is a knowledge-augmented document search task jointly proposed by KAIST and Samsung, which identifies semantic targets in documents through a single natural query in real-time. Unlike traditional machine reading comprehension, KTRL+F incorporates external knowledge embeddings into phrase embeddings, balancing speed and performance. The model achieves accurate and comprehensive searching and retrieval within documents by enhancing contextual knowledge, aiming to improve information access efficiency. The project focuses on utilizing external knowledge to identify semantic targets in documents through natural queries in real-time, providing an enhanced search experience.