This　paper　investigates　two　kinds　of　convergence　behaviors　of　nonlinear　polar　opinion　dynamics　over　cooperative-antagonistic　social　networks　determined　by　signed　graphs.　The　first　problem　studies　the　neutralization　of　opinions　over　strongly　connected　and　structurally　unbalanced　signed　graphs,　while　the　second　problem　focuses　on　the　containment　convergence　to　external　stubborn　agents　over　signed　graphs　under　a　variety　of　connectivity　constraints.　It　is　observed　that　various　connectivity　conditions　induce　the　common　intrinsic　property　that　the　involved　generalized　opposing　Laplacian　matrix　is　nonsingular.　A　stability　framework　is　proposed　that　addresses　both　the　existence　of　potential　limiting　behaviors　as　well　as　their　convergence　properties.　This　involves　an　analysis　of　the　positiveness　of　the　(opinion)　susceptibility　functions.　The　domains　of　attraction　associated　with　the　convergent　behaviors　are　characterized.　The　solvability　of　both　problems　is　provided,　and　the　attracting　domains　are　characterized　precisely.　The　results　are　applied　to　blended　social　networks　composed　of　three　specialized　opinion　models.　Simulations　and　comparisons　are　provided　to　illustrate　the　effectiveness　of　the　paper＇s　results.　©　2025　Elsevier　Ltd