With　the　rapid　development　of　sensing　hard-devices,　wireless　communication　technologies　and　smart　mobile　devices,　a　large　number　of　data　for　moving　objects　have　been　collected,　among　which　a　group　of　high　precision　data　(e.g.,　GPS)　are　widely　used　for　traffic　predictions　and　management.　However,　in　modern　city　life,　a　large　volume　of　positioning　data　of　moving　objects　is　collected　with　low-precision　positions,　which　causes　the　difficulty　for　trajectory　match,　analysis　or　group　identification.　In　view　of　this　limitation,　this　paper　proposes　a　novel　method　for　the　semantic　trajectory　based　group　identification.　Specifically,　the　trajectory　data　are　used　to　discover　the　spatial　and　semantic　information　of　persons　to　calculate　their　similarities.　Based　on　which,　the　groups　of　persons　with　strong　correlations　are　identified.　To　evaluate　our　method,　we　conduct　several　experiments　on　Geolife　dataset.　The　experimental　results　show　that　the　proposed　method　has　a　significant　effect　on　the　group　identification.