This document describes the DCIM approach for mining closed frequent itemsets (CFIs) in massively distributed environments. It discusses: 1) The DCIM algorithm which uses MapReduce to distribute the mining of CFIs across multiple nodes. It counts item frequencies in the first pass, then mines CFIs using a prime number approach to avoid costly operations. 2) Experimental results on Wikipedia and ClueWeb datasets showing DCIM has significantly better performance than other parallel CFI mining methods, with speedups increasing as dataset size and minimum support threshold decrease. 3) The authors conclude DCIM is an efficient and effective parallel algorithm for CFI mining in big data due to its data modeling using prime