Cyber　social　computing　has　brought　great　changes　and　potential　intelligent　technologies　for　wireless　networks.　Among　these　technologies,　information　caching　strategies　are　promising　approaches　to　achieving　lower　delay,　higher　throughput　and　energy　efficiency　(EE)　of　user　equipment　(UE)　in　5G　wireless　networks,　by　deploying　intelligent　caching　and　computing　at　the　mobile　edge.　However,　the　static　information　caching　strategies　ignore　the　relevance　of　traffic　fluctuation　among　different　base　stations　(BSs)　and　the　variance　of　users＇　interests.　Thus　in　this　paper,　an　information　caching　strategy　for　cyber　social　computing　based　wireless　network　is　proposed,　taking　advantages　of　two　layer　social　cyberspaces　in　both　traffic　correlation　between　BSs　and　the　social　relationship　between　UEs.　In　the　first　layer,　a　base　station　social　network　(BSSN)　is　constructed　based　on　the　social　relationship　between　BSs,　which　is　defined　as　social-tie　factor　(STF).　In　the　second　layer,　the　Indian　Buffet　Model　(IBM)　is　used　to　describe　the　social　influence　of　one　UE　to　another.　To　reduce　base　station＇s　traffic　load,　users　with　similar　social　interest　can　share　the　contents　they　cached　with　each　other.　Therefore,　device-to-device　(D2D)　communication　is　taken　as　the　underlay　to　cellular　networks　in　our　proposed　information　caching　strategy.　By　utilizing　the　social　characteristic　of　BSSN,　the　very　important　BSs　(VIBSs)　with　higher　averages　STF　are　selected.　Then　the　normal　small　cells　(NSCs)　within　the　VIBS＇s　coverage　are　linked　to　the　VIBSs　only　and　the　other　unique　small　cells　(USCs)　will　be　routed　back　into　the　core　network　(CN)　directly.　Limited　cache　and　backhaul　capacity　in　the　whole　network　are　only　shared　by　VIBSs　and　USCs.　UEs　will　communicate　with　each　other　via　D2D　links　only　if　they　have　i)　similar　interests,　ii)　enough　encounter　duration　between　users　and　iii)　are　adjacent　with　each　other.　Otherwise,　the　UE　shall　obtain　the　required　contents　via　cellular　networks.　With　the　tool　of　stochastic　geometry　process,　key　performance　indicators,　e.g.,　coverage　probability,　network　throughput　power　consumption,　EE,　delay　and　offloaded　traffic　are　studied.　Both　the　theoretical　and　numerical　results　show　that　the　proposed　information　caching　strategy　for　cyber　social　computing　can　achieve　high　coverage　probability,　throughput,　EE　and　delay　by　optimizing　STF　threshold,　VIBS　coverage　and　D2D　communication　radius.