Automated Service Recommendation Using Deep Network for Augmented Feature Vector

In today's digitally driven era there is an enormous growth in the usage of online services and to make this online journey effortless for the consumers, recommending suitable services is a must. Our model Deep Modified Matrix Factorization for Augmented Feature Vector(DeepMMF) helps in automating an efficient recommendation service for the consumers. Our model DeepMMF uses an Augmented Feature Vector which is embedded by using both explicit and implicit information for users and items that identify the correlation between them. DeepMMF uses deep neural networks with a modified matrix factorization technique which finds the similarities between the users by decomposing the highly sparse user-item interaction matrix into three low-dimensional matrices. This matrix captures the most significant latent factors representing the users and items and thus recording the user preferences. Our model outperforms the other state-of-the-art approaches for the benchmark real-world dataset Movielens 1M, 10M, 20M, and 25M.