Location Based Personalized Recommendation Model Using Optimized Recurrent Neural Network

The e-Commerce industries require efficient recommendation model to maximize the profit and user satisfaction. Recommendation system assists consumers to find their relevant items of interest. The state-of-art models are designed by considering long-term consumer context. However, in the current application dynamic, such long-context does not exist and recommendation must be made based on user present behavior of an ongoing session. Many session based approaches have been presented in recent times to forecast user's nextitem requirement. However, these models consider modeling a single behavior with long-context. As a result, the state-of-art model finds difficulty in revealing the correlation between the items and behaviors. To overcome the above-mentioned research challenge the research work presents a multi-behavioral trait based on consumer location-centric prediction (LCP) model using an optimized recurrent neural network (ORNN). LCP model can learn both short and long-context efficiently. Experiment outcome shows LCP attain significant performance over the existing model in terms of Recall, F1-Score, Mean reciprocal rate (MMR) and Hit rate (HR)