Let quantitativo represent an unknown document and let y represent per random target author’s stylistic ‘profile’. During one hundred iterations, it will randomly select (a) fifty a cent of the available stylistic features available (ed.g. word frequencies) and (b) thirty distractor authors, or ‘impostors’ from a pool of similar texts. Sopra each iteration, the GI will compute whether incognita is closer preciso y than to any of the profiles by the thirty impostors, given the random selection of stylistic features con that iteration. Instead of basing the verification of the direct (first-order) distance between quantitativo and y, the GI proposes puro record the proportion of iterations sopra which quantita was indeed closer esatto y than to one of the distractors sampled. This proportion can be considered a second-order metric and will automatically be verso probability between niente and one, indicating the robustness of the identification of the authors of incognita and y. Our previous sistema has already demonstrated that the GI system produces excellent verification results for classical Latin prose.31 31 Complice the setup durante Stover, et supporto naughtydate al, ‘Computational authorship verification method’ (n. 27, above). Our verification code is publicly available from the following repository: This code is described con: M. Kestemont et al. ‘Authenticating the writings’ (n. 29, above).
For modern documents, Koppel and Winter were even able puro report encouraging scores for document sizes as small as 500 words
We have applied per generic implementation of the GI sicuro the HA as follows: we split the individual lives into consecutive samples of 1000 words (i.e. space-free strings of alphabetic characters), after removing all punctuation.32 32 Previous research (see the publications mentioned con the previous two taccuino) suggests that 1,000 words is verso reasonable document size durante this context. Each of these samples was analysed individually by pairing it with the profile of one of the HA’s six alleged authors, including the profile consisting of the rest of the samples from its own text. We represented the sample (the ‘anonymous’ document) by verso vector comprising the incomplete frequencies of the 10,000 most frequent tokens con the entire HA. For each author’s profile, we did the same, although the profile’s vector comprises the average imparfaite frequency of the 10,000 words. Thus, the profiles would be the so-called ‘mean centroid’ of all individual document vectors for verso particular author (excluding, of course, the current anonymous document).33 33 Koppel and Seidman, ‘Automatically identifying’ (n. 30, above). Note that the use of per single centroid verso author aims onesto scampato, at least partially, the skewed nature of our giorno, since some authors are much more strongly represented sopra the campione or background pool than others. If we were not using centroids but mere text segments, they would have been automaticallysampled more frequently than others during the imposter bootstrapping.
Preciso the left, verso clustering has been added on apice of the rows, reflecting which groups of samples behave similarly
Next, we ran the verification approach. During one hundred iterations, we would randomly select 5,000 of the available word frequencies. We would also randomly sample thirty impostors from verso large ‘impostor pool’ of documents by Latin authors, including historical writers such as Suetonius and Livy.34 34 See Appendix 2 for the authors sampled. The pool of impostor texts can be inspected durante the code repository for this paper. Con each iteration, we would check whether the anonymous document was closer preciso the current author’s profile than onesto any of the impostors sampled. Con this study, we use the ‘minmax’ metric, which was recently introduced con the context of the GI framework.35 35 See Koppel and Winter, ‘Determining if two documents’ (n. 26, above). For each combination of an anonymous text and one of the six target authors’ profiles, we would primato the proportion of iterations (i.ancora. verso probability between nulla and one) con which the anonymous document would indeed be attributed puro the target author. The resulting probability table is given con full mediante the appendix to this paper. Although we present verso more detailed tete-a-tete of this giorno below, we have added Figure 1 below as an intuitive visualization of the overall results of this approach. This is verso heatmap visualisation of the result of the GI algorithm for 1,000 word samples from the lives con the HA. Cell values (darker colours mean higher values) represent the probability of each sample being attributed esatto one of the alleged HA authors, rather than an imposter from verso random selection of distractors.