Interactive digital humans can serve as narrators or guides in smart exhibition applications, enhancing personalization, engagement, and immersion. We argue that for a talking digital character, the co-speech gestures are not just optional complements to verbal responses but essential components for creating human-like and immersive interaction experiences. To validate this idea, we implement a system capable of reenacting synchronized gestures for new speeches in a simulated smart exhibition application. This method constructs gesture motion graphs from different reference data and can provide co-speech gestures with distinctive styles much more rapidly compared to typical data-driven methods. The results show that our system succeeds in reenacting co-speech gestures with the specified style, and the reenacted co-speech gestures are preferred by human evaluators in multiple aspects compared to the baseline setups.