Authors: Armel Jacques Nzekon Nzeko'O ^1,2,3; Maurice Tchuenté ^1,2; Matthieu Latapy ³

• 1 Unité de modélisation mathématique et informatique des systèmes complexes [Bondy]
• 2 Centre d’Excellence Africain en Technologies de l’Information et de la Communication
• 3 ComplexNetworks

Recommending appropriate items to users is crucial in many e-commerce platforms. One common approach consists in selecting the N most relevant items for each user. To achieve this, recom-mender systems rely on various kinds of information, like item and user features, past interest of users for items and trust between users. Current systems generally use only one or two such pieces of information, which limits their performance. In this paper, we design and implement GraFC2T2, a general graph-based framework to easily combine various kinds of information for top-N recommendation. It encodes content-based features, temporal and trust information into a graph model, and uses personalized PageRank on this graph to perform recommendation. Experiments are conducted on Epinions and Ciao datasets, and comparisons are done with systems based on matrix factorization and deep learning using F1-score, Hit ratio and MAP evaluation metrics. The results show that combining different kinds of information generally improves recommendation. This shows the relevance of the proposed framework.