Multi-Modal Analysis of Presentation Media Data

With the explosive development of multimedia information systems, the fusion analysis of the multi-modal information is a research trend. Inspired by this, this study proposes a presentation media data processing algorithm based on multi-modal analysis, which mainly includes a multi-modal processing framework for image data and video data. This study combines scene graph generation, tensor fusion, user label hyper-graph and other modeling ideas to build a new multi-modal analysis paradigm to solve the problems in multi-modal analysis. For image multi-modal analysis, this study proposed a two-stage prediction model based on scene graph generation, which performs modal feature analysis through modified tensor fusion. For video multi-modal analysis, this study designed a new collaborative filtering algorithm based on user label hyper-graph, which comprehensively considers video activity and the user behavior weight to optimize recommendation effect. More specifically, targeting the dissemination characteristics of demonstration media data, this study finally analyzed the information transmission mode of human-computer combination and designed a dissemination optimization method based on multi-modal information processing. By using data from three public datasets and generating test cases, the experimental results of this study show that this method performs well in terms of accuracy, robustness and other indicators.