The　vanilla　influence　maximization　problem　requires　some　kind　of　seeds　before　the　diffusion　process　so　as　to　maximize　the　expected　influence　spread　in　a　social　network.　This　problem　has　been　extensively　studied　due　to　its　applications　in　viral　marketing.　However,　most　studies　require　selecting　all　seeds　at　once,　which　wastes　part　of　the　budget　due　to　not　utilizing　the　observation　results.　This　paper　considers　adaptive　influence　maximization　and　adaptive　stochastic　influence　maximization　problems　under　a　general　feedback　model,　where　seeds　can　be　selected　after　a　fixed　number　of　observation　time-steps.　Generally,　the　objective　function　lacks　the　adaptive　submodularity　property,　making　it　difficult　to　construct　effective　approximate　solutions.　We　introduce　a　comparative　factor　and　present　a　theoretical　analysis　of　the　solution　using　an　adaptive　greedy　framework　to　solve　them.　In　addition,　a　feasible　approximation　algorithm　based　on　the　reverse　sampling　technique　is　used　to　solve　the　adaptive　stochastic　influence　maximization　problem.　©　2021,　Springer　Nature　Switzerland　AG.