This document discusses a proposed algorithm for community detection in dynamic social networks. It involves applying multi-resolution techniques to a multi-objective immune algorithm. The algorithm aims to maximize community quality and minimize temporal cost. It has three modules: 1) calculating modularity and betweenness values, 2) identifying high similarity vertex pairs, and 3) regrouping isolated vertices based on modularity values. A case study on Facebook is provided to demonstrate detecting strong and weak communities based on user activities like photos tagged, comments, and posts shared. The algorithm is presented as the first phase for community detection in dynamic networks, with the second phase still under development.