Model-Based Classification With Constraints: Who Wrote Each Chapter In Finbar’s Hotel?

Finbar’s Hotel is a multi-author book which was written by seven contemporary Irish authors. A novelty aspect of the book is that the names of the seven authors are known, but the assignment of authors to the seven chapters is unknown; each chapter has only one author. A model-based classification procedure is developed to determine the authorship of each chapter of the book. Vocabulary distributions are modelled using Sichel distributions which are estimated using text samples from the chapters and from books by the seven authors. Constraints are introduced into the model-based classification procedure so that a one-to-one matching of chapters and authors is achieved. A comparison of the maximum likelihood and Bayesian approaches to implementing the model-based classification procedure is completed. Finally, we give an answer to the question “Who wrote each chapter of Finbar’s Hotel?”