Distributed Algorithm for Text Documents Clustering Based on k-Means Approach

The presented paper describes the design and implementation of distributed k-means clustering algorithm for text documents analysis. Motivation for the research effort presented in this paper is to propose a distributed approach based on current in-memory distributed computing technologies. We have used our Jbowl java text mining library and GridGain as a framework for distributed computing. Using these technologies we have designed and implemented k-means distributed clustering algorithm in two modifications and performed the experiments on the standard text data collections. Experiments were conducted in two testing environments-a distributed computing infrastructure and on a multi-core server.