Research And Implementation Of Social Relationship Network Construction Based On Large Scale Video Data

As the continuous development of multimedia technology,video has become the main form for people to obtain information.Extracting social relationships from large scale video data and building social networks can quickly and effectively mine deep semantic information in large-scale video data.In this paper,the construction method of social network in large scale video data is studied and explored,mainly including the following three aspects.First,a video character social relationship extraction method(TSM)based on semantic objects is proposed to solve the problem of lack of semantic object features in current research.Firstly,abundant video feature representation is obtained by extracting spatio-temporal features and semantic object information.Then,the propagating knowledge map is used to capture the interaction between semantic objects and video scenes.Finally,attention mechanism is introduced to measure the effectiveness of each semantic object in different scenes.A large number of comparative experiments verify the accuracy of TSM model in the task of extracting social relations of video characters.Second,a parallel construction method of video character social network based on Spark is proposed.In order to solve the problem of the lack of edge attributes in the construction of video social network,a method of network construction considering both edge weights and attributes is proposed.The edge attributes and weights are assigned through the results of the extraction of social relations of characters.In order to solve the challenge of social network construction in large scale video data,a parallel strategy of network construction process is proposed,which parallelizes the video social network construction process from multiple granularity.Finally,the performance of parallel network construction model is verified by experiments.Thirdly,the algorithms and functions proposed in this paper are integrated into big data analysis platform(BDAP)system to expand the functions related to video processing in the system.The related components are designed in detail,and the video social relationship extraction and social network building components and their sub modules are tested to verify the functional effectiveness and availability of the integrated components of the platform.