Hosts in mobile social networks (MSN), in many applications, see each other in repetitive places at relatively specific times. Awareness of such contact regularity and formed communities can help to improve the efficiency of routing protocols in this network. We propose the Artemia router that takes advantage of this knowledge. In order to know the contact regularity, this router divides each day into specific intervals and registers the presence of hosts on a daily, weekly, and monthly basis with relevant bit maps. In addition, it detects and registers communities in each interval with a multi-criteria optimization approach. A modeled utility function is proposed to forward messages in this router. In that model, we present a low-complexity solution that uses Chebyshev's inequality to increase the accuracy of the data to identify community, friend, familiar, and stranger nodes. Artemia improves the network overhead by up to 44% and 35%, and the delivery rate and delivery latency metrics by about 6% and 8%, respectively, compared to two state-of-the-art protocols, Dynamic Routing and TCCB.