The uniqueness of keystroke data: Benefits of analyzing input data in cases of questioned authorship.

摘要

The study of questioned authorship, wherein analysts attempt to determine the most likely author for a given text, can be approached with a variety of tools, such as stylistic, stylometric, and computational assessments. A recent addition to this analytic arsenal are keystroke loggers, programs that record every keystroke to provide a "real-time" record of the writing process. Keystroke logging adds to our understanding of an author's idiolectal style, the personal preferences in discourse production that distinguish one author from another. Therefore, this research seeks to determine in what ways do authors individuate themselves through their writing processes? Using a case study of ten participants, this study examines such keystroke features as typos, corrections, and timing to determine which features, if any, can be used to identify an author's unique writing style. While manner of correction and to a lesser extent timing data were helpful in determining authorship, an individual's use of function keys proved to be immediately and immensely valuable for the analysis.