Classification of Web Documents using Fuzzy Logic Categorical Data Clustering

We propose a categorical data fuzzy clustering algorithm to classify web documents. We extract a number of words for each thematic area (category) and then, we treat each word as a multidimensional categorical data vector. For each category, we use the algorithm to partition the available words into a number of clusters, where the center of each cluster corresponds to a word. To calculate the dissimilarity measure between two words we use the Hamming distance. Then, the classification of a new document is accomplished in two steps. Firstly, we estimate the minimum distance between this document and all the cluster centers of each category. Secondly, we select the smallest of the above minimum distance and we classify the document in the category that corresponds to this distance.