In　an　uncertain　environment,　the　process　of　fuzzy　cross-evaluation　in　data　envelopment　analysis　(DEA)　models　is　explained　by　different　types　of　fuzzy　data.　In　this　paper,　the　fuzzy　cross-efficiency　matrix　with　different　types　of　fuzzy　data　will　be　changed　into　a　cross-evaluation　similarity　matrix　by　similarity　theory　of　generalized　trapezoidal　fuzzy　numbers.　Based　on　the　fuzzy　information　retrieval　(FIR)　method,　a　new　ordered-geometric-mean　averaging　operator　is　defined　to　yield　aggregation　weights　by　the　relevance　between　the　self-evaluation　and　cross-evaluation　of　each　decision-making　unit　(DMU).　Furthermore,　the　common　neighbor　DMU　selection　algorithm　is　constructed　to　classify　the　DMUs.　Then,　the　fuzzy　logic　rules　are　proposed　to　modify　the　aggregation　results　by　combining　the　clustering　results.　Finally,　this　paper　demonstrates　the　applicability　of　fuzzy　cross-efficiency　aggregation　using　the　FIR　method　by　taking　an　example　of　12　suppliers　from　the　semiconductor　industry.