Support Vector Machines for Large Scale Text Mining in R

SVM are an established tool in machine learning and data analysis. Though many implementations of SVM exist often specific applications require tailor made algorithms. In text mining in particular the data often comes in large sparse data matrices. Typical SVM algorithms like SMO do not take advantage of the sparsity, and do not scale well to data sets with millions of entries. In this paper we present an implementation of linear SVM’s for R that address both of these issues.