As　an　important　research　direction　in　the　complex　social　network,　the　difficulty　of　community　detection　lies　in　the　search　and　discovery　of　social　structures　efficiently　and　accurately.　In　this　study,　an　algorithm　named　SEOCD　(Seeds　Extension　Overlapping　Community　Detection)　for　overlapping　community　detection　based　on　random　walk　and　seeds　extension　is　proposed　in　order　to　solve　the　problem　of　seeds　selection　and　expansion　in　many　seed-based　algorithms.　First,　SEOCD　uses　the　random　walk　strategy　to　find　the　seed　communities　with　tight　structures.　Second,　from　the　seed　communities,　the　similarity　between　each　pair　of　node　and　community　is　calculated.　The　nodes　whose　similarity　greater　than　a　predefined　threshold　are　selected.　Third,　The　strategy　of　optimizing　a　self-adaptive　function　is　used　to　expand　the　communities.　Finally,　The　free　nodes　in　the　network　are　assigned　to　their　corresponding　communities,　which　finds　out　all　the　overlapping　community　structures.　Experiments　on　real　and　artificial　networks　show　that　SEOCD　is　capable　of　discovering　overlapping　communities　in　complex　social　networks　efficiently.　©　2017　Association　for　Computing　Machinery.