Abstract: With the development of IT technology, customer trading system is developing rapidly, and plays an increasingly important role in real life. Along with the trading number growing, using customer trading behaviors and historical trading data, a well decision support could be gained from system administrator. The algorithm of best cluster mining could find a most closely associated sub-graph from a graph structure. The most closely associated customers could be effectively found from customer trading system by way of algorithm GG-LSH, thus some effective decision supports are provided for people.

Key words: best cluster mining, sub-graph, algorithm GG-LSH, decision support