Complete syntactic N-grams as style markers for authorship attribution

Juan Pablo Posadas-Duran; Grigori Sidorov; Ildar Batyrshin

doi:10.1007/978-3-319-13647-9_2

Complete syntactic N-grams as style markers for authorship attribution

Juan Pablo Posadas-Duran, Grigori Sidorov, Ildar Batyrshin

Producción científica: Contribución a una revista › Artículo › revisión exhaustiva

18 Citas (Scopus)

Resumen

In this paper we present an authorship attribution method based on the use of complete (non-continuous, with bifurcations) syntactic n-grams as style markers. Syntactic n-grams are obtained by following paths in subtrees of a syntactic tree. We work with relatively short text fragments and build authors’ profiles of various sizes using tf-idf scheme. We train SVM classifier to perform the task. We compare the method with the application of character n-grams and show that the accuracy increases when using complete syntactic n-grams.

Idioma original	Inglés
Páginas (desde-hasta)	9-17
Número de páginas	9
Publicación	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volumen	8856
DOI	https://doi.org/10.1007/978-3-319-13647-9_2
Estado	Publicada - 2014

Acceder al documento

10.1007/978-3-319-13647-9_2

Otros archivos y enlaces

Enlace a la publicación en Scopus

Citar esto

@article{fd0cfe11485a4c7d81dd71ec45713e7a,

title = "Complete syntactic N-grams as style markers for authorship attribution",

abstract = "In this paper we present an authorship attribution method based on the use of complete (non-continuous, with bifurcations) syntactic n-grams as style markers. Syntactic n-grams are obtained by following paths in subtrees of a syntactic tree. We work with relatively short text fragments and build authors{\textquoteright} profiles of various sizes using tf-idf scheme. We train SVM classifier to perform the task. We compare the method with the application of character n-grams and show that the accuracy increases when using complete syntactic n-grams.",

keywords = "Authorship attribution, SVM, Style markers, Syntactic markers, Syntactic n-grams, Syntactic paths",

author = "Posadas-Duran, {Juan Pablo} and Grigori Sidorov and Ildar Batyrshin",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2014.",

year = "2014",

doi = "10.1007/978-3-319-13647-9_2",

language = "Ingl{\'e}s",

volume = "8856",

pages = "9--17",

journal = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

issn = "0302-9743",

publisher = "Springer Verlag",

}

Complete syntactic N-grams as style markers for authorship attribution. / Posadas-Duran, Juan Pablo ; Sidorov, Grigori ; Batyrshin, Ildar.
En: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 8856, 2014, p. 9-17.

Producción científica: Contribución a una revista › Artículo › revisión exhaustiva

TY - JOUR

T1 - Complete syntactic N-grams as style markers for authorship attribution

AU - Posadas-Duran, Juan Pablo

AU - Sidorov, Grigori

AU - Batyrshin, Ildar

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2014.

PY - 2014

Y1 - 2014

N2 - In this paper we present an authorship attribution method based on the use of complete (non-continuous, with bifurcations) syntactic n-grams as style markers. Syntactic n-grams are obtained by following paths in subtrees of a syntactic tree. We work with relatively short text fragments and build authors’ profiles of various sizes using tf-idf scheme. We train SVM classifier to perform the task. We compare the method with the application of character n-grams and show that the accuracy increases when using complete syntactic n-grams.

AB - In this paper we present an authorship attribution method based on the use of complete (non-continuous, with bifurcations) syntactic n-grams as style markers. Syntactic n-grams are obtained by following paths in subtrees of a syntactic tree. We work with relatively short text fragments and build authors’ profiles of various sizes using tf-idf scheme. We train SVM classifier to perform the task. We compare the method with the application of character n-grams and show that the accuracy increases when using complete syntactic n-grams.

KW - Authorship attribution

KW - SVM

KW - Style markers

KW - Syntactic markers

KW - Syntactic n-grams

KW - Syntactic paths

UR - http://www.scopus.com/inward/record.url?scp=84921341475&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-13647-9_2

DO - 10.1007/978-3-319-13647-9_2

M3 - Artículo

AN - SCOPUS:84921341475

SN - 0302-9743

VL - 8856

SP - 9

EP - 17

JO - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

JF - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -

Complete syntactic N-grams as style markers for authorship attribution

Resumen

Acceder al documento

Otros archivos y enlaces

Huella

Citar esto