Service　selection　is　the　key　technology　during　the　fusion　of　AmI　and　Service　Computing,　the　selection　procedure　is　influenced　by　service＇s　QoS,　user　preference,　Context　and　other　factors　in　Am!　system.　According　the　features　of　AmI,　the　paper　proposes　a　two　phase　interaction　decision　making　model　which　combines　the　service＇s　objective　attributes　with　user＇s　subjective　preference.　It　computes　objective　weight　of　the　candidate　services＇　QoS　attribute　values　by　using　their　information　entropy,　then　the　user　can　reference　for　the　objective　decision　information　and　give　his　subjective　selection　information　which　includes　preference　and　Context-aware　constraints.　At　last,　the　optimized　selection　rule　is　computed　by　using　the　objective　and　subjective　information.　The　test　results　show　that　the　solution　has　fine　adaptability　and　usability,　and　it　achieves　the　unity　of　generality　and　personalized.