The　existing　user　authentication　technology　can　easily　judge　the　legitimacy　of　the　user＇s　identity　entering　the　system,　but　it　can　not　prevent　the　user　with　legal　authorization　from　malicious　damage　to　the　system.　In　order　to　prevent　the　above　situation　from　bringing　huge　losses　to　enterprises　or　organizations　and　evaluate　whether　the　user　behavior　of　internal　users　is　credible,　this　paper　proposes　a　user　behavior　credibility　measurement　method.　This　method　constructs　the　user　behavior　graph　around　the　user　behavior　timing　information,　and　constructs　the　user　behavior　credibility　measurement　model　based　on　the　user　behavior　graph　combined　with　PageRank　algorithm.　The　model　uses　parallel　genetic　strategy　to　calculate　the　user　credibility　threshold　under　spark　framework.　The　experimental　results　show　that　the　efficiency　of　genetic　algorithm　is　significantly　improved　compared　with　that　of　traditional　single　machine　and　Hadoop　platform,　and　it　also　has　an　obvious　effect　on　the　accuracy　of　measurement　model.　©　2021　IEEE.