Abstract |
In this report we present the results of evaluating the speed of
text mining experiments using the OpenMosix cluster software. Our primary purpose is to develop a computational environment where the time taken to perform compute intensive experiments over a large range of parameter settings is minimized. This will lead to clearer analysis of algorithms and ultimately to a better choice of models based on empirical results. Results are shown that characterize each of the cluster components and also show the speedup in performance over compute nodes. |