Correlation Clustering - Minimizing Disagreements on Arbitrary Weighted Graphs

We solve several open problems concerning the correlation clustering problem introduced by Bansal, Blum and Chawla [1]. We give an equivalence argument between these problems and the multicut problem. This implies an O(logn) approximation algorithm for minimizing disagreements on weighted and unweighted graphs. The equivalence also implies that these problems are APX-hard and suggests that improving the upper bound to obtain a constant factor approximation is non trivial. We also briefly discuss some seemingly interesting applications of correlation clustering. There is a correlation between the creative and the screwball. So we must suffer the screwball gladly. Kingman Brewster, Jr. (1919–1988) President Yale University (1963–1977), US Ambassador to Great Britan (1977-1981), Master of University College, London (1986-1988).