Taxi　is　an　important　part　of　urban　public　transportation　which　meets　the　demands　of　special　people　to　travel　from　door　to　door.　Taxi　trip　characteristics　are　influenced　by　districts　location　and　travel　purposes.　This　paper　extracts　the　time　and　location　information　of　taxi　alighting　based　on　the　taxi　meter　data　and　taxi　GPS　data.　Based　on　the　point-of-interest　(POI)　and　searching　popularity　of　the　POI　from　online　map　data,　this　paper　also　utilizes　the　semantic　reasoning　methods　to　predict　the　purpose　of　taxi　travels,　and　taxi　trips　are　grouped　into　three　most　popular　types:　commuting　travel,　business　travel　and　external　travel.　Then,　multi-source　data　analysis　models　are　proposed　to　calculate　the　characteristic　parameters　of　taxis　trips　including　average　travel　mileage,　travel　time,　regional　travel　intensity　of　three　types　of　taxi　trips.　The　case　study　of　Beijing　selects　three　typical　areas　of　different　land　use　types.　The　analysis　result　shows　that　trip　characteristics　of　different　areas　are　various　and　the　travel　distance　in　resident　area　is　shortest,　and　the　travelers　leave　for　external　transportation　hub　usually　have　lower　sensitivity　to　travel　mileage;　and　the　taxi　travel　mileage　of　business　areas　are　almost　same　with　resident　areas,　however　the　travel　time　is　more　longer.　The　analysis　results　of　taxi　trips　characteristics　of　different　areas　revealed　in　this　paper　provide　significant　reference　for　acquiring　the　taxi　travel　demand　and　travel　characteristics,　the　taxi　stations　planning,　the　estimation　of　the　reasonable　amount　of　the　taxis　and　the　operation　of　the　intelligent　Taxi　Dispatching　System　(TDS).