The rapid expansion of the mobile Internet has led to online social networks becoming an increasingly integral part of our daily lives, this offers a new perspective in the study of human behavior. User' interest groups are now considered crucial elements within social networks. The objective of event evolution and tracking within social networks is the ability to monitor the real-time evolution of user' interests based on the previous diffusion behavior of influence disseminators and to anticipate future diffusion behavior of users. However, prior studies have only addressed the evolution regarding a single interest or the evolution of user interest in the entire interest space, ultimately with no modeling of user interest evolution within event tracking. In order to address these challenges, this study proposes a knowledge-infused deep learning-based event tracking model named DIEET (Diffusion and Interest Evolution behavior modeling for Event Tracking). This model accurately predicts the propagation and interest evolution behavior in event tracking by considering both propagation and interest evolution behavior. Specifically, the DIEET model incorporates the interval time, the number of times, the sequence interval time, and finally user preference for the event of interest, greatly improving the accuracy and efficiency of event evolution prediction. The experiments conducted on real Twitter datasets detail the proposed DIEET models' ability to greatly improve the tracking of the state of user interest alongside the popularity of event propagation, and DIEET also has superior prediction performance compared to state-of-the-art models in terms of identifying user dynamic interest. Therefore, the aforementioned model offers promising potential in the ability for predicting and tracking the evolution of user interest and event propagation behavior on online social networks.