An Approach to Detecting Text Autorship in the Spanish Language

Authors tend to express themselves using language in ways that reflect particular styles, vocabularies, biases, idioms, etc. These features can be captured in the so-called firm or stylone. Although capturing these attributes with high fidelity has proven to be very challenging, some advances have been made. Stylometry is the analysis of the unique attributes that are expressed by an author unconsciously through his or her publications. In this paper we investigate techniques for the detection of authorship patterns from the text content of a large number of digital documents, including e-mails, academic notes and free redaction in the Spanish language. A mechanism based on pondering parameters, including statistical observations, extracting a pattern is proposed. We defined 150 stylistic criteria parameters adapted to the Spanish language to compute our metric. Extensive experiment results are also presented.