A Novel Document Clustering Algorithm Using Squared Distance Optimization Through Genetic Algorithms

摘要

K-Means Algorithm is most widely used algorithms in document clustering. However, it still suffer some shortcomings like random initialization, solution converges to local minima, and empty cluster formation. Genetic algorithm is often used for document clustering because of its global search and optimization ability over heuristic problems. In this paper, search ability of genetic algorithm has exploited with a modification from the general genetic algorithm by not using the random initial population.A new algorithm for population initialization is given in this paper and results are compared with k-means algorithm.