With　development　of　the　infrastructures　supporting　smart　home　which　has　entered　in　a　new　stage　featured　by　intelligent　services.　The　services　based　on　natural　language　instructions　are　aimed　at　simplifying　and　improving　our　lives.　To　customize　and　develop　these　services　more　efficiently,　this　paper　proposes　an　approach　to　model　and　execute　services　based　on　natural　language　instructions　at　runtime　which　introduces　the　knowledge　graph　into　development　process.　Firstly,　a　concept　model　of　knowledge　graph　is　introduced　for　smart　home　services.　Secondly,　we　put　forward　the　mechanism　to　construct　the　instance　of　knowledge　graph　for　smart　home　services.　Finally,　rules　of　transformation　are　provided　to　achieve　mapping　from　natural　language　instructions　to　the　services　of　knowledge　graph.　We　evaluate　our　model　on　a　prototype　system　and　the　experimental　results　show　that　our　approach　can　develop　services　at　runtime　and　effectually　reduce　the　lines　of　code　by　90%.　©　2019　Knowledge　Systems　Institute　Graduate　School.　All　rights　reserved.