Depicting User Connections in Social Media by Mapping User-Shared Images

Millions of images are publicly available on various social networking sites due to their sharing behaviour. Enormous amount of such images have been collected and are trained using multiple machine learning algorithms and machine generated tags to obtain user relationships. Images have been obtained and analysed from multiple online social networking sites. These images are passed into a trained image classifier model which is based on Convolutional Neural Network (CNN) using Tensor-Flow. The obtained result is further clustered and segregated by analysing the data to obtain a behaviour in order to study the various relationship patterns between the users. Using the information obtained, a graphical representation of the studied pattern is formulated. This proposed engine is proven to be 47 percent better than other approaches.