In　the　process　of　software　development,　people　pay　more　and　more　attention　to　the　trustworthiness　attribute　of　software,　and　the　trustworthiness　description　of　software　requirements　has　become　an　important　part　of　requirements　engineering.　For　the　description　of　trusted　requirements,　most　of　the　methods　still　use　the　description　of　words　and　graphics,　and　their　contents　are　limited　to　the　category　of　specific　business　logic,　which　fails　to　abstract　some　highly　universal　contents.　Under　the　trend　of　more　and　more　complex　trusted　requirements,　it　not　only　causes　the　increase　of　requirements　understanding　cost　and　the　waste　of　resources　to　repeat　the　existing　work　in　the　requirements　process,　but　also　leads　to　the　result　that　the　requirements　data　is　not　easy　to　be　mined　and　utilized.　In　order　to　improve　this　situation　(purpose),　this　paper　proposes　a　new　model　construction　and　description　method.　On　the　basis　of　previous　research,　this　paper　constructs　the　trusted　requirements　and　behavior　model　of　software　based　on　ontology　formal　modeling　method,　and　describes　the　differences　between　functions　by　taking　the　trusted　concerns　of　software　requirements　as　the　characteristic　attributes　of　requirements　in　different　scenarios;　Finally,　the　formal　description　is　used　as　the　data　source,　and　the　process　of　expecting　behavior　from　requirements　is　completed　according　to　the　attributes　in　ontology　and　the　random　forest　algorithm　based　on　bagging　idea.　This　method　can　help　developers　to　better　reuse　the　past　experience　in　the　process　of　requirements　acquisition,　and　can　more　reasonably　and　accurately　anticipate　the　behavior　of　software.　©　2021　IEEE.