To　restrain　the　spread　of　harmful　information　is　crucial　for　the　healthy　and　sustainable　development　of　social　networks.　We　address　the　problem　of　restraining　the　spread　of　harmful　information　by　immunizing　nodes　in　the　networks.　Previous　works　have　developed　methods　based　on　the　network　topology　or　studied　how　to　immunize　nodes　in　the　presence　of　initial　infected　nodes.　These　static　methods,　in　which　nodes　are　immunized　at　once,　may　have　poor　performance　in　the　certain　situation　due　to　the　dynamics　of　diffusion.　To　tackle　this　problem,　we　introduce　a　new　dynamic　immunization　problem　of　immunizing　nodes　during　the　process　of　the　diffusion　in　this　paper.　We　formulate　the　problem　and　propose　a　novel　heuristic　algorithm　by　dealing　with　two　sub-problems:　(1)　how　to　select　a　node　to　achieve　the　best　immunization　effect　at　the　present　time?　(2)　whether　the　selected　node　should　be　immunized　right　now?　Finally,　we　demonstrate　the　effectiveness　of　our　algorithm　through　extensive　experiments　on　various　real　datasets.　©　2018　Elsevier　B.V.