Abstract: Private information retrieval techniques can protect user query privacy while conducting data retrieval. However, the existing privacy information retrieval protocols are still not efficient enough and have relatively weak functions. Based on homomorphic encryption and number theory principles, an efficient ciphertext matching mechanism is proposed, and a non-interactive symmetric private data retrieval protocol is designed to support fuzzy queries. From the perspectives of efficiency and functionality, the efficiency of privacy information retrieval is significantly improved, and the functionality of the protocol is expanded. Specifically, in the proposed scheme, a single query on tens of thousands of records only requires a delay of seconds, superior to the most efficient keyword-based privacy information retrieval schemes currently available in terms of server-side latency. Additionally, this scheme allows for fuzzy queries with wildcards and multi-keyword queries, enabling users to obtain an experience similar to plaintext queries while protecting user and data privacy.