In　recent　years,　Xinjiang　CPI　has　more　impacts　on　CPI　of　China,　thus　accuracy　of　the　CPI　Prediction　of　Xinjiang　has　vital　significance　to　China.　On　the　basis　of　the　construction　of　theoretical　framework　of　internet　search　data　and　Xinjiang　CPI,　using　the　Baidu　keyword　recommendation　technology　automatically　this　paper　obtained　more　than　300　appropriate　keywords,　then　it　synthesized　keywords　and　formed　a　composite　index　by　method　of　stepwise　synthesis,　the　fitting　degree　of　the　synthetic　index　was　99.57%.　And　then,　it　empirically　analyzed　relationship　between　Xinjiang　CPI　and　composite　index　of　internet　search.　The　empirical　results　showed　that　changes　in　the　composite　index　of　network　search　can　be　a　good　indicator　in　the　fluctuations　of　Xinjiang　CPI,　the　mean　absolute　percentage　error　in　the　prediction　was　2.60%.