Figure 1.

System architecture: Phase ${\phi }_1$ is a pre-classification used for feature extraction (top left, machine learning process); Phase ${\phi }_2$ uses the classification results and the features to build the base retail ontology (middle and top right blocks); Phase ${\phi }_3$ derives the evolved ontology after adding the new information (center part); Phase ${\phi }_4$ uses NCF (middle left block) jointly with CF and CBF (top right) to build the ensembled recommendations.