Abstract
We define the partial weighted set cover problem, a generic combinatorial optimization problem, that includes some classical data mining problems as special cases. We prove that it is computationally intractable and give a local search algorithm for this problem. As application examples, we then show how to translate clustering and outlier detection problems into this generic problem. Our experiments on synthetic and real-world datasets indicate that the quality of the solution produced by the generic local search algorithm is comparable to that obtained by state-of-The-Art clustering and outlier detection algorithms.