In the era of Internet and big data, fragmented learning has become the predominant mode of learning. However, the fragmented nature of knowledge content makes it difficult for individuals to build an efficient knowledge network. Existing knowledge management tools often fall short in terms of knowledge association and systematization, failing to meet users’ needs for knowledge integration. This study proposes a fragmented knowledge management behavior model based on connectivism theory, aimed at helping learners effectively manage and integrate scattered knowledge resources. By constructing the fragmented knowledge management behavior model, the study identifies the needs and behavioral characteristics of learners at different stages of knowledge management, emphasizing the importance of the relationships and connections between knowledge for improving learning efficiency. Based on this, an intelligent personal knowledge management system was designed, which utilizes knowledge graphs and knowledge chain visualization techniques to help users better understand the relationships between pieces of knowledge and achieve effective integration and efficient management. Usability testing results show that the system performs well in improving learning efficiency, optimizing knowledge integration, and enhancing user experience. This study provides new theoretical and practical references for the design of fragmented knowledge management systems.