Implicit Feedback Recommendation for Plant-pollinator Networks

Examining and evaluating complex ecological communites is a popular study area of significant ecological value. Mutualistic plant-pollinator networks are an especially important community because of the crucial role of pollinator in food production, yet much is still unkown about the underlying community structures of these networks, as well as their stability under enviornmental pertubation. We present a method adapting recommender system collaborative filtering algorithms for the plant-pollinator network in an attempt to quantify pollinator preferences and tendancies and predict future behavior. We apply an Implicit Feedback Matrix Factorization (IFMF) model originally developed for a TV show recommender engine and discuss why this application is ecologically viable. Our results indicate that IFMF significantly outperforms a random-prediction model and is ripe for further devolopment.