¹Instituto Venezolano de Investigaciones Científicas, Departamento de Matemáticas, Estado Miranda, Venezuela. Estudiante de postgrado. Email: freddy.vate01@gmail.com

En este artículo se aplican algunos métodos de clasificación a un conjunto de textos con el objetivo de estudiar la probabilidad que el libro Novela de la tía fingida haya sido escrita por Miguel de Cervantes. Esta novela se le ha atribuido históricamente, pero existen algunas posiciones encontradas al respecto. Los métodos usados en este artículo contemplan: regresión logística, regresión logística aditiva, análisis discriminante lineal, cuadrático, regularizado, de mezclas y flexible, árboles de clasificación, método de los k-ésimos vecinos más cercanos, método de Bayes ingenuo y máquinas de soporte vectorial.
Los métodos fueron calibrados y aplicados utilizando un corpus de autores contemporáneos a Cervantes (Lope de Vega, Jerónimo de Pasamonte, Alonso Fernández de Avellaneda, Mateo Alemán y Francisco de Quevedo) junto con más de cuarenta variables, principalmente palabras y signos de puntuación, medidas sobre muestras de los textos escritos por estos autores.
Con respecto a estos métodos, la mayoría clasifica la obra como cervantina; sin embargo, es recomendable ampliar el corpus utilizado para el estudio e incluir más autores para la comparación.

Palabras clave: análisis discriminante, árboles de clasificación, máquinas de aprendizaje, regla de Bayes, regresión logística, validación cruzada.

In this paper, some classification methods are applied to a set of texts with the aim of studying the probability that the book Novela de la tía fingida has been written by Miguel de Cervantes. This novel has been historically attributed to him but there are some encountered positions about this. The methods used in this paper range from: logistic regression, additive logistic regression, linear, quadratic, regularized, mixture and flexible discriminant analysis, classification tree, k-nearest neighbour, Naive Bayes method and support vector machines.
Methods were trained and applied using a corpus of authors contemporary to Cervantes as Lope de Vega, Jerónimo de Pasamonte, Alonso Fernández de Avellaneda, Mateo Alemán, and Francisco de Quevedo and more than forty variables, mainly words and punctuation marks, measured over written texts by these authors.
Respect to these methods, most of them classify the novel as another Cervantes work; however, is our recommendation to include more texts from these authors and more authors.

Key words: Bayes rule, Classification tree, Cross validation, Discriminant analysis, Logistic regression, Machine learning.

Texto completo disponible en PDF

Referencias

1. Aylward, E. T. (1982), Cervantes: Pioneer and Plagiarist, Tamesis Books Limited, Londres, UK.         [ Links ]

2. Baum, L. F. (2001), The Royal Book of Oz, Dover Publications, New York, States United. Escrito con colaboraci\'on' de R. Thompson.         [ Links ]

3. Binongo, J. (2003), `Who Wrote the 15th Book of Oz? An Application of Multivariate Analysis to Authorship Attribution´, Chance 16(2), 9-17.         [ Links ]

4. Bird, S., Klein, E. & Loper, E. (2009), Natural Language Processing with Python, O'Really, Sebastopol, States United.         [ Links ]

5. Dempster, A. P., Laird, N. M. & Rubin, D. B. (1977), `Maximum likelihood from incomplete data via the EM algorithm´, Pattern Recognition 39, 1-38.         [ Links ]

6. Gardner, M. (1998), Visitors from Oz: The Wild Adventures of Dorothy, the Scarecrow, and the Tin Woodman, St Martins Press, New York, States United.         [ Links ]

7. Grieve, J. (2007), `Quantitative Authorship Attribution: an Evaluation of Techniques´, Literacy and Linguistic Computing 22(3), 251-270.         [ Links ]

8. Hastie, T., Tibshirani, R. & Friedman, J. (2009), The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2 edn, Springer, New York, States United.         [ Links ]

9. Hoover, D. L. (2002), `Multivariate Analysis and Study of Style Variation´, Literacy and Linguistic Computing 18(4), 341-360.         [ Links ]

10. Hosmer, D. & Lemeshow, S. (2000), Applied Logistic Regression, 2 edn, Wiley, New York, States United.         [ Links ]

11. Jockers, M., Witten, D. & Criddle, C. (2008), `Reassessing authorship of the Book of Mormon using delta and nearest shrunken centroid classification´, Literacy and Linguistic Computing 23(4), 465-491.         [ Links ]

12. Johnson, R. & Wichern, D. (1998), Applied Multivariate Statistical Analysis, Fourth edn, Prentice Hall, New York, States United.         [ Links ]

13. Jolliffe, I. T. (2002), Principal Component Analysis, 2 edn, Springer, New York, States United.         [ Links ]

14. Joula, P. (2006), `Authorship Attribution´, Foundations and Trends in Information Retrieval 1(3), 233-334.         [ Links ]

15. Koppel, M., Schler, J. & Argamon, S. (2009), `Computational methods in authorship attribution´, Journal of the American Society for Information Science and Technology 60(1), 9-26.         [ Links ]

16. Lebart, L., Morineau, A. & Warwick, K. (1984), Multivariate Descriptive Statistical Analysis, John Wiley & Sons, New York, States United.         [ Links ]

17. Madrigal, J. L. (2003), `De c\'omo y por qu\'e La t\'\ia fingida es de Cervantes´, Artifara(2).         [ Links ]

18. Rencher, A. (2002), Methods of Multivariate Analysis, Second edn, Wiley, New York, States United.         [ Links ]

19. Ripley, B. D. (1996), Pattern Recognition and Neural Networks, Cambridge University Press, Cambridge, UK.         [ Links ]

20. Tibshirani, R., Hastie, T., Narashimhan, B. & Chu, G. (2003), `Class prediction by nearest shrunken centroids with applications to DNA microarrays´, Statistical Science 18(1), 104-117.         [ Links ]

21. Venables, W. N. & Ripley, B. D. (2002), Modern Applied Statistics with S, Fourth edn, Springer, New York, States United. *http://www.stats.ox.ac.uk/pub/MASS4         [ Links ]

22. Witten, I. H. & Frank, E. (2005), Data Mining: Practical Machine Learning Tools and Techniques, 2 edn, Elsevier, San Francisco, States United.         [ Links ]

23. Yu, B. (2008), `An evaluation of text classification methods for literacy studies´, Literacy and Linguistic Computing 23(3), 327-343.         [ Links ]

Este artículo se puede citar en LaTeX utilizando la siguiente referencia bibliográfica de BibTeX: