Large-scale　group　decision　making　(LSGDM)　is　characterized　by　a　large　number　of　participators,　multiple　groups　and　a　mass　of　decision　data　provided　by　the　participators.　With　the　development　of　social　media　and　e-democracy　technologies,　the　social　relationships　among　group　decision　makers　should　be　taken　into　consideration　when　we　address　LSGDM　problems　with　incomplete　fuzzy　preference　relations.　In　this　paper,　a　social　network　community　detection　approach　of　social　networks　based　on　the　fuzzy　clustering　method　is　proposed.　Then,　a　method　of　repairing　incomplete　fuzzy　preference　relations　based　on　the　divided　communities　is　proposed.　Moreover,　we　have　proposed　a　two-stage　consensus　reaching　method　to　balance　the　number　of　iterations　and　modification　range　of　original　decision　information,　in　which　the　group　closeness　centrality　of　community　is　applied　to　measure　its　importance.　Also,　the　procedure　for　LSGDM　with　fuzzy　preference　relations　based　on　social　network　community　analysis　is　established.　The　feasibility　and　advantages　of　the　proposed　method　for　LSGDM　based　on　social　network　community　analysis　are　illustrated　by　an　illustrative　example　of　a　flexible　manufacturing　system.