The purpose of this study is to design effective working systems which adapt to changes in human needs by developing an algorithm which forms workers into optimal groups using the meausre of cohesion. Three major results can be derived from the study. Firstly, the algorithm developed here provides an optimal point at which to stop clustering. Secondely, the entropic measure of cohesion having an internal probabilistic structure is superior with respect to any other methods proposed before as far as the design of workgroup is concerned. Thirdly, the r $C_{n}$ clustering algorithm is better than the dichotonomic one.e.