Deep Learning-Based Triphase Community Detection for Multimedia Data

Abstract

Network analysis plays a significant role in business which is achieved through community detection. The relationship between the nodes is mined by community detection, which facilitates the analysis of complicated networks. In the current era, social media is the commonest mode of communication that leads to complex online social networks, from which useful information can be retrieved. Not only textual data is shared in social media, in addition multimedia data plays a significant role in content sharing in social network. This research work proposes a community detection method based on deep learning algorithm to detect communities for both textual content and multimedia content. The proposed method triphase–deep learning community detection (TriDL-CD) method illustrates the relationship of data in a graphical way in the first step. Second step converts the graph into user relationship table using similarity weightage analysis from which the communities are formed using convolution neural network. The proposed method proved to be efficient in detecting high quality communities for multimedia content compared with Louvain community detection algorithm, Leiden community detection, and surprise community detection algorithms. Experimental results show the efficacy of deep learning in the concept of community detection.