This study investigates the sliding mode control for semi-Markov jump systems with dynamic self-triggered protocol. Utilizing an average dwell time strategy, the variation of semi-Markov chain is regulated by a deterministic switching signal. To improve transmission efficiency and achieve better control performance, a dynamic self-triggered protocol is developed, which predicts the next triggered time based on the current sampled data, obviating the need for dedicated hardware to continuously monitor system states and eliminating the Zeno phenomenon. A sliding mode controller related to the self-triggered protocol is proposed, which ensures the finite-time reachability. Based on the proposed self-triggered protocol and Lyapunov theory, a sufficient criterion is established to guarantee mean-square exponential stability under partial unknown transition rates. Finally, the effectiveness of the proposed method is verified through simulation examples.