A Statistical Methodology for Writer Verification1

A statistical methodology for automatically determining whether a pair of documents, a known and a questioned, were written by the same individual is proposed. First, several writing elements, e.g., characters, are extracted from each document. Second, differences between corresponding elements from each document are computed. Third, using conditional probability estimates of each difference, the log-likelihood ratio (LLR) is computed for the hypotheses that the documents were written by the same writer and by different writers; the conditional probability estimates are determined in a learning phase from labelled writing samples using either Gaussian mixtures or Parzen window estimates. Fourth, the LLR is analyzed with the Tippett plot to provide a measure of the strength of evidence.