Community-based recipe recommendation and adaptation in peer-to-peer networks

In this paper, we propose a distributed recipe recommendation mechanism that utilizes social information for adaptive recipe recommendation in a peer-to-peer (P2P) network. A recipe flavor model is first proposed for modeling recipes and validating recipe adaptations. Peers in the network group themselves into communities in which members share common preferences of recipe data. This is helpful to visit more relevant peers when the query scope is fixed thus improve the performance of recommendation. Based on a graph-based recipe representation, we propose a recipe similarity measure and a filtering algorithm to generate candidates of cooking recipes to be recommended. A recipe adaptation method is also proposed in order to better match users' preferences. Simulation experiments are conducted for the evaluation of the proposed model and the results show good performance of it.