Knowledge extraction methods from document collections
Argyris Kalogeratos · 2013
dataset using k-sp variants. . . . . . . . . .80 4.4 The NMI, Purity measures for the refined solutions found for each dataset.Bold values indicate the best result per column.The underlined t-values denote the cases where according to the statistical t-test k-sp appears not to be significantly better (0<t-val<1.999), or appears to be worse than the compared method (t-val<0). . . . . . . . . . . . . . . . . . . . . . . . . . .85 5.1 Different parameter setups that reduce the generic clustering framework procedure to popular incremental algorithms. . . . . . . . . . . . . . . . .107 6.1 Results for synthetic datasets with fixed k * =20 clusters with 200 datapoints in each cluster. . . . . . . . . . . . . . . . . . . .