Faced　with　a　large　number　of　the　efficiency　evaluation　problems　with　non-homogeneous　decision-making　units　(DMUs)　in　reality,　the　existing　efficiency　evaluation　methods　are　always　unfair　in　terms　of　evaluation　perspective　and　inputs/outputs　allocation.　Therefore,　this　paper　adopts　the　concept　of　network　data　envelopment　analysis　(DEA)　to　define　the　inclusion　relationship　of　production　structure　between　non-homogeneous　DMUs.　Sequentially,　the　DEA　cross-efficiency　model　and　its　corresponding　cross-evaluation　strategy　are　constructed　by　combining　the　self-evaluation　perspective　with　peer-evaluation　perspective,　and　then　the　efficiency　of　these　DMUs　can　be　evaluated　in　a　more　objective　and　reasonable　way.　Actually,　the　cross-evaluation　perspective　can　fully　reflect　the　effect　of　non-homogeneous　structure　on　efficiency.　In　addition,　the　inclusion　relationship　is　extended　to　a　mixed　relationship,　and　the　application　scope　of　the　new　cross-efficiency　method　is　thus　further　expended.　Finally,　two　examples　are　provided　to　illustrate　the　effectiveness　of　the　new　method.