Dynamic knowledge-based recommendation system for E-commerce using deep neural network
Last few decades Internet is playing very crucial role in development of online business.It provides a very good platform for both buyers as well as sellers. It also creates a huge volume of online information.These information are very valuable for upgradation of the online businesses, but it becomes a very mediokres task to process these information. The concepts of data mining and artificial intelligence are playing a vital role to overcome these problems by creating efficient recommendation models. Collaborative filtering is one of the best techniques used in the development of such systems, but cold start, sparsity and scalability are the main problems that degrade the quality of the system. Hence, proposed model has used web usage mining, artificial intelligence and deep neural networks(DNN) concepts to develop new recommendation system. It has two phases, computation of ratings of the items using web usage mining, and then the computation of prediction ranking scores of these items using DNN for feature representation. Finally, it has been compared with various renouned models like KaF, NCF, SVD and PMF using two public datasets that verified the performance of the proposed system using the Mean Absolute Error (MAE) evaluating parameter.