Word　segmentation　is　a　basic　topic　in　the　field　of　natural　language　processing,　and　improving　the　accuracy　of　word　segmentation　is　a　key　problem.　With　the　popularity　of　microblog,　accurate　word　segmentation　for　microblog　text　has　become　a　hot　spot.　However,　microblog　texts　often　contain　information　about　multiple　related　domains,　ambiguous　words　in　multi-domain　will　lead　to　the　decline　of　word　segmentation　accuracy.　Based　on　the　model　theory　of　word　vector　and　branching　entropy,　this　paper　proposes　a　multi-domain　global　correlation　degree　branching　entropy　method　for　microblog　text　word　segmentation.　This　model　is　applied　to　microblog　text　about　house　price　topic　in　Beijing.　The　precision,　recall　and　F-measure　of　this　method　are　compared　with　branching　entropy　model　proposed　by　Zhang[6],　and　the　experimental　results　show　that　our　method　outperforms　it.　©　2020　ACM.