In　software　engineering,　the　undirected　compound　graph　serves　as　a　medium　for　human-machine　communication,　translating　human　intent　into　machine　language　to　efficiently　deliver　and　demonstrate　people’s　demands　by　utilizing　its　outstanding　analysis　capability.　Still,　the　industrial　adoption　of　goal　models　is　hindered　by　the　scalable　data　and　the　massive　manual　workload　it　incurs.　Overt　problems,　including　the　uncertainty　of　iteration　direction　and　the　difficulty　of　controlling　the　number　of　edge　crossings,　remain　unsolved.　We　introduce　the　method　of　graph　folding,　which　continuously　collapse　indecisive　nodes　to　reduce　the　data　scale　and　remain　the　basic　structure　of　the　original　graph　simultaneously　to　provide　a　possible　solution　to　this　dire　state.　The　folding　method　confirms　the　direction　of　the　iterations　and　considerably　reduces　its　number,　which　greatly　improves　the　efficiency　of　the　man-machine　communication.　We　repeatedly　operate　adjustment　and　expansion　of　the　graph　to　ensure　the　optimal　solution,　and　the　results　are　promising.　©　2023,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.