Published Paper Details:

The Event Based Social Network(ESBN) is a distributed platform for organizing the online social events, in which events are submitted to the user groups on associated topic, location and time period. Usually, event based social network systems contain abundant events and large scale of user participating on event, there exist an overhead on clustering the user for the event. In order to model a participant prediction, deep learning architecture for the user recommendation has become necessary. However, many existing work using machine learning model leads to scalability and sparsity issues. Further events and user profiles continually arrive and evolve in terms of behaviour and knowledge, which leads to the issues of task cold-start. In order to overcome the above mentioned challenge, a novel dense recurrent neural network architecture using deep neural network for participant recommendation on basis of evolution of user knowledge and preference has been formulated in this research. Initially proposed idea deals with the dynamics of the user and event in the event based social network by extracting the features with multiple latent factors using feature extraction technique. Latent factor extracted on basis of the event and user features according to their experience and historical behaviors. Further deep neural network has been incorporated to model a participant recommendation system to event on basis of latent feature extracted towards prediction on timely basis to achieve high reliability, accuracy and latency. Deep learning architecture schedules the participants to event on embedding latent features on generation of the objective function to produce recommendation with minimum error. It results in significant increase on the prediction performance for discriminative information's of the event and user. Extensive experiments have been conducted on real datasets to compare proposed model with conventional. The experimental results show that deep learning architecture can achieve both effectiveness and good scalability on large scale data.