Abstract: Movie data is a multimodal data set containing three different modes of data,which are movies,directors and actors.It is challenging how to extract features from these multi-modal data and reveal the relationships and influences of features within and among different modes through appropriate visualizations and interactions.In this paper,we first extract the content,numerical and relational data features from corresponding movie data set and design the proper visual layouts according to different characteristics of data features.Based on Martini vitreous narrative visual structure we design the visual analysis system to explore the multi-modal movie data.Especially for the filmmakers’relationship,we define the cooperation between different modes of filmmakers,and an improved force-directed graph layout algorithm is proposed to visualize the cooperative relationships between filmmakers.The Web-based multi-modal movie data visualization system MDVis has been implemented and used to visualize and analyze the Douban movie data set.The experimental results verify the effectiveness of our method in analyzing multi-modal movie data.