Self-Adaptive Recommendation Systems: Models and Experimental Analysis
Luca Becchetti, Ugo Maria Colesanti, Alberto Marchetti-Spaccamela, Andrea Vitaletti · 2008
We design and study recommendation algorithms for a fully decentralized scenario in which each item/node of a network recommends other items/nodes only on the basis of simple statistics on the behavior of users that visited the node in the past. We perform a theoretical and experimental study assessing that very simple heuristics can provide recommendations of good quality even in such a restrictive scenario.