To　support　a　large　amount　of　GPS　data　generated　from　various　moving　objects,　the　back-end　servers　usually　store　low-sampling-rate　trajectories.　Therefore,　no　precise　position　information　can　be　obtained　directly　from　the　back-end　servers　and　uncertainty　is　an　inherent　characteristic　of　the　spatio-temporal　data.　How　to　deal　with　the　uncertainty　thus　becomes　a　basic　and　challenging　problem.　A　lot　of　researches　have　been　rigidly　conducted　on　the　uncertainty　of　a　moving　object　itself　and　isolated　from　the　context　where　it　is　derived.　However,　we　discover　that　the　uncertainty　of　moving　objects　can　be　efficiently　reduced　and　effectively　ranked　using　the　context-aware　information.　In　this　paper,　we　focus　on　context-aware　information　and　propose　an　integrated　framework,　Context-Based　Uncertainty　Reduction　and　Ranking　(CURR),　to　reduce　and　rank　the　uncertainty　of　trajectories.　Specifically,　given　two　consecutive　samplings,　we　aim　to　infer　and　rank　the　possible　trajectories　in　accordance　with　the　information　extracted　from　context.　Since　some　context-aware　information　can　be　used　to　reduce　the　uncertainty　while　some　context-aware　information　can　be　used　to　rank　the　uncertainty,　to　leverage　them　accordingly,　CURR　naturally　consists　of　two　stages:　reduction　stage　and　ranking　stage　which　complement　each　other.　We　also　implement　a　prototype　system　to　validate　the　effectiveness　of　our　solution.　Extensive　experiments　are　conducted　and　the　evaluation　results　demonstrate　the　efficiency　and　high　accuracy　of　CURR.