Flexible recommender systems based on graphs

Joseph Pellegrino (1)

1 Kernix Software

— The objective of this paper is to share Kernix's approach to build flexible recommender systems based on graph oriented databases. The integration of entities, and relationships between them, into a unique graph structure allows to design recommendations as graph traversals. This approach offers a flexible framework allowing to handle the variety of entities of interest and enabling to design rich strategies in order to compute recommendations for various use cases.

Mots clés: Graph database; Recommender systems; Data Management

Domaines: Réseaux sociaux et d'information [cs.SI]; Intelligence artificielle [cs.AI]

Joseph Pellegrino. Flexible recommender systems based on graphs. AISR2017, May 2017, Paris, France. ⟨hal-01640313⟩