The　proliferation　of　smart　devices　has　accelerated　the　development　of　mobile　applications　(Apps).　Traffic　generated　by　Apps　accounts　for　the　vast　majority　of　cellular　traffic,　resulting　in　significant　transmission　delay.　Caching　popular　App　content　can　effectively　reduce　duplicate　data　transmission.　In　this　paper,　we　design　an　App-based　cooperative　content　caching　scheme　through　Apps＇　traffic　prediction.　In　detail,　a　tensor　domain　adaptation　network　model　for　Apps＇　traffic　prediction　is　proposed,　where　cross-domain　spatio-temporal　features　are　transferred　from　the　source　domain　to　the　target　domain　by　learning　a　shared　tensor　subspace.　Furthermore,　a　cooperative　content　caching　strategy　is　proposed　to　minimize　the　service　delay.　The　cache　decision　problem　is　solved　by　GA-PSO　which　jointly　uses　a　genetic　algorithm　and　a　particle　swarm　optimization.　Experiment　results　show　the　proposed　prediction　model　outperforms　existing　models　in　terms　of　prediction　accuracy,　and　the　performance　impact　of　different　caching　modes　and　storage　sizes　is　also　Investigated.　©　2022　IEEE.