Hybrid recommendation system and methods based on collaborative filtering integrated with multiple neural networks

Abstract

A hybrid recommendation system may generate recommendations, predictions, and/or classifications by applying collaborative filtering to influence Convolutional Neural Networks ('CNNs'), Recurrent Neural Networks (“RNNs”), and/or other neural networks that model characteristic, structural, sequential, contextual, interactive, and/or other relationships from interactions of different users. The system may provide different user interactions as input to a first neural network, and the first neural network may model relationships between the different users and different items based on the interactions. The system may track activities of one or more users, may use a personalized model, that is generated via collaborative filtering of the tracked activities, together with other models of the relationships to generate a recommendation matrix, and may modify a user interface to present recommended candidate items based on recommendation matrix vectors that rank the recommended candidate items higher than other items.