In this paper we have address that dimension selection,dimension weighting and data assignment are three essential tasks for high dimensional data clustering. We also pointed out that constraints are necessary to break the circular-dependency of the three essential tasks. We proposed the integrated constraint based high dimensional data
clustering (ICBC) algorithm which,unlike previous algorithms, uses constraint to accomplish all the
three essential tasks. Experimental results have shown the super- iority of integrated usage of
constraints in all the three essential tasks in terms of accuracy, efficiency and scalability.
Future work includes exploiting more information from constraints and to cooperating them with
search methods of other subspace clustering to improve the performance of high dimensional data
clustering.