Local　community　detection　is　an　innovative　method　to　mine　cluster　structure　of　extensive　networks,　that　can　mine　the　community　of　seed　node　without　the　need　for　global　structure　information　about　the　entire　network,　as　distinct　from　the　global　community　detection　algorithm,　it　is　efficient　and　costs　less.　However,　a　key　problem　with　this　field　is　that　the　location　of　seed　nodes　affects　the　performance　of　the　algorithm　to　a　great　extent,　and　it　is　easy　to　add　abnormal　nodes　to　the　community,　the　robustness　of　the　algorithm　is　low.　In　this　study,　we　proposed　a　novel　algorithm　named　CAELCD.　First,　find　the　high-quality　seed　node　of　the　community　starting　from　the　initial　seed　node,　so　as　to　avoid　the　seed-dependent　problem.　Second,　generate　the　community＇s　core　area　and　expand　to　get　the　local　community,　which　solves　the　problem　that　the　expansion　from　a　single　seed　prefers　to　add　the　wrong　nodes.　Experiments　on　the　parameter,　accuracy　and　visualization　of　the　CAELCD　are　designed　on　networks　with　different　characteristics.　Experimental　results　demonstrate　that　CAELCD　has　superior　performance　and　high　robustness.