The　field　of　attributed　graph　clustering　has　garnered　increasing　attention,　particularly　with　the　advent　of　graph　convolutional　network　(GCN),　which　have　deepened　our　understanding　of　learning　both　attribute　and　structural　information　in　graphs.　Existing　graph　deep　embedding　clustering　methods　typically　learn　attribute　or　structural　information　alone,　or　integrate　attribute　information　into　a　learning　network　for　structural　information.　However,　these　methods　fail　to　fully　integrate　the　available　information.　Therefore,　we　propose　a　novel　deep　attributed　graph　clustering　method　named　Attention-based　Graph　Clustering　Network　with　Dual　Information　Interaction　(ADIIN).　Specifically,　an　attention-based　interaction　fusion　module　is　presented　to　adaptively　incorporate　two　types　of　information　and　propagate　the　fused　information　to　both　networks　interactively.　Additionally,　it　can　adjustively　integrate　information　from　each　hidden　layer　at　different　scales　based　on　attentional　mechanisms.　Furthermore,　we　design　a　more　robust　quadruple　joint　self-supervision　strategy　to　align　node　attribute　representation,　linear　fusion　representation,　and　multi-scale　feature　fusion　representation,　thereby　enhancing　the　clustering　performance　of　the　entire　model.　Extensive　experiments　conducted　on　several　benchmark　datasets　demonstrate　that　our　proposed　method　outperforms　state-of-the-art　deep　clustering　methods.　Our　code　is　publicly　available　at　https://github.com/sliboo/ADIIN.　©　2025　Elsevier　B.V.