Faced with the plight of declining visitor traffic, major museums are actively seeking new changes and exploring more attractive personalised experience designs. In this paper, we take “demand-optimisation” as the research idea to analyse the personalised demand for museum visit experience. Under the guidance of experience design thinking, the solution for a digital roaming museum service platform with a community function has been designed. In the optimization path chain of “information acquisition - self-creation - memory experience”, the digital museum reconstruction combining hierarchical modelling and octree, the A* algorithm based virtual interactive roaming, the social interaction based virtual roaming, and the social experience based virtual interactive roaming are proposed respectively. The digital museum reconstruction with hierarchical modelling and octree, the virtual interactive roaming based on A* algorithm, and the souvenir recommendation based on social network analysis are proposed, respectively. The experimentally constructed digital museum scene has fewer collisions and a higher frame rate, which can have a more comprehensive information perception on the basis of fully restoring the real experience. Meanwhile, the interactive roaming test shows that the paths generated by the algorithms in this paper are fitted smoothly and can meet the personalized roaming needs in the virtual museum scene. In addition, the MAE of the recommendation result for the souvenir guide module is only 0.012, with a high degree of matching. In terms of scene information display, roaming freedom, and souvenir recommendation suitability, the above optimized link networks enhance the museum visiting experience.