Читать книгу Methodologies and Challenges in Forensic Linguistic Casework - Группа авторов - Страница 33

 The susceptibility of authorship analysis to contextual bias could be explored experimentally. i.e. telling authorship analysts alternative stories around a set problem, and seeing if they come up with different answers. This could be applied to both stylistic and computational approaches to authorship analysis.

 It would be useful to explore ways in which computational and more qualitative approaches to authorship analysis might be combined. e.g. using heavily computational methods to elicit a large set of features but to also to examine those features to rule out ones without linguistic explanation.