Link-Based Clustering Algorithm for Clustering Web Documents

Clustering web documents involves the use of a large amount of words to be inputted to clustering algorithms such as K-Means, Cosine Similarity, Latent Discelet Allocation, and so on. This causes the clustering process to consume much time as the number of words in each document increases. In many web documents, web links are available along with the contents; these web link texts may contain a tremendous amount of information for clustering. In our work, we show that just using the web link text alone gives better clustering efficiency than considering the whole document text. We implemented our algorithm with two benchmark datasets, and the results show that the clustering efficiency is increased by our algorithm more than the existing methods.