In　the　study　of　cloud　computing　and　its　applications,　security　has　been　regarded　as　one　of　the　greatest　problems　in　the　development　of　cloud　computing.　Therefore,　the　study　and　prevention　of　computer　virus　in　cloud　computing　infrastructure　becomes　practically　significant　for　cloud　security.　Based　on　the　characteristics　of　the　rapid　spread　of　computer　viruses,　when　a　community　which　is　composed　by　cloud　nodes　has　a　higher　influence　in　the　cloud-based　network,　it　has　a　greater　potential　hazard　if　there　are　some　viruses　lurking　in　the　community.　Therefore,　it＇s　necessary　to　find　the　influential　community　for　virus　prevention　in　cloud　infrastructure.　To　evaluate　cross-community　influence,　this　paper　proposes　a　novel　method　that　consists　of　two　components:　(1)　the　Mutual　Evaluation　Learning　(MEL)　model　for　evaluating　the　pairwise　influence　between　two　nodes　(clients,　servers)　and　(2)　a　cross-community　influence　ranking　algorithm　(CCIR)　designed　to　quantify　the　influential　strength　of　each　cross-community　based　on　MEL　and　PageRank.　Empirical　studies　on　numerous　real　networks　demonstrate　that　the　proposed　method　can　adapt　to　different　networking　scenarios　and　can　reasonably　and　effectively　reflect　the　distribution　of　community　influence.