In　Ambient　Intelligence　(AmI)　environment,　data　access,　data　storage,　data　processing,　and　data　presentation　are　much　different　from　what　they　are　in　the　the　traditional　computing　environment.　These　issues　bring　some　new　obstacles　and　challenges　to　the　system　developing　for　AmI　According　to　these　challenges,　this　paper　proposed　a　data　model,　as　well　as　a　data　management　model　＂AmI-Data＂　for　the　AmI　space.　AmI-Data　supports　the　automatic　hoarding　and　harvesting　of　data　from　multiple,　heterogeneous　data　sources.　It　also　supports　metadata　to　give　the　system　the　ability　of　service　discovery　between　heterogeneous　AmI　systems.　Furthermore,　AmI-Data　also　supports　a　lot　of　advanced　data　mining　methods,　such　as　HMM,　ANN,　Fuzzy　computing.　The　final　data　presentation　of　the　AmI-Data　is　＂service＂,　which　is　the　final　goal　of　Ambient　Intelligence　applications.