Aiming at the problems of low user satisfaction and response efficiency existing in current research results, this paper proposes a method of extracting learning resources of digital library based on personalized push service. Collect data such as user resource download information, access URL content and system access time, select small log file parsing tool to directly analyze the collected data, and save the analyzed data to MySQL or Oracle database. According to the data collection and processing results of personalized behavior of readers, fuzzy clustering is carried out for visiting users of digital library by using the clustering method under the equivalence matrix, and personalized push service is provided for users by using the optimal clustering of browsing, borrowing and resource utilization. The fuzzy recognition method is used to identify the individual situation of target users, and the retrieval resources are fed back to users according to the recognition results to achieve the extraction of learning resources. The experimental results show that the extraction method of learning resources in the digital library has better performance, higher user satisfaction, faster response time and higher extraction accuracy, showing good application performance.