Output details

A new unsupervised feature selection method for text clustering based on genetic algorithms

<17>Feature selection is an important step in text mining that has been explored mainly by use of unsupervised selection methods that define measures for evaluating the discriminative power of every single term. In contrast, this research has devised a measure to evaluates a group of terms. Because of the huge search-space for finding the most valuable group of terms, a GA has been devised for reaching the final feature vector. Experimental results confirm that this method can outperform current state of the art methods on different corpuses.