The recent expansion and development of digital information resources has significantly changed the function of academic libraries from that of a repository of printed material to that of a source of knowledge. The many choices of information, however, can create problems for the user in trying to find the information they are looking for, efficiently. The personalized Information Services (PIS), with Machine Learning (ML)- based recommender systems, are a promising solution to overcome the issue. Such systems employ algorithms to process the user's preferences, search patterns, borrowing history, and academic interests to make personalized recommendations. The article discusses how to use machine learning to improve personalized information services in academic libraries. It summarizes literature, analyzes the state-of-theart recommender system approaches, outlines a digitization framework and shares the expected outcomes. The results show that the use of the ML-based recommender systems has a significant impact on information discovery, user satisfaction, resource usage, and research productivity. This framework is designed to leverage user profiling, data analysis, recommendation algorithms, and ongoing feedback systems for intelligent library services. This research also adds to the increasing body of work in the field of smart libraries, offering valuable insights into the implementation and advantages of machine learning personalized libraries.