A method for calculating term similarity on large document collections | IEEE Conference Publication | IEEE Xplore