Search Grouping: unlikely collision implementation

Develop an implementation that is more efficient when the number of group values is very large and hence multiple items per group is less likely.