POS Tagging without a Tagger: Using Aligned Corpora for Transferring Knowledge to Under-Resourced Languages

Autores/as

  • Ines Turki Khemakhem University of Sfax, MIRACL Laboratory
  • Salma Jamoussi University of Sfax, MIRACL Laboratory
  • Abdelmajid Ben Hamadou University of Sfax, MIRACL Laboratory

DOI:

https://doi.org/10.13053/cys-20-4-2430

Palabras clave:

POS tagging, alignment, parallel corpus, under-resourced languages.

Resumen

Almost all languages lack sufficient resources and tools for developing Human Language Technologies (HLT). These technologies are mostly developed for languages for which large resources and tools are available. In this paper, we deal with the underresourced languages, which can benefit from the available resources and tools to develop their own HLT. We consider as an example the POS tagging task, whichis among the most primordial Natural Language Processing tasks. The task is important because it assigns to word tags that highlight their morphological features by considering the corresponding contexts. The solution that we propose in this research work, is based on the use of aligned parallel corpus as a bridge betweena rich-resourced language and an under-resourced language. This kind of corpus is usually available. The rich-resourced language side of this corpus is annotated first. These POS-annotations are then exploited topredict the annotation on the under-resourced language side by using alignment training. After this training step, we obtain a matching table between the two languages, which is exploited to annotate an input text. The experimentation of the proposed approach is performed for a pair of languages: English as a rich-resourced language and Arabic as an under-resourced language. We used the IWSLT10 training corpus and English TreeTagger. The approach was evaluated on the test corpus extracted from the IWSLT08 and obtained Fscoreof 89%. It can be extrapolated to the other NLP tasks.

Biografía del autor/a

Ines Turki Khemakhem, University of Sfax, MIRACL Laboratory

Researcher andassistant at Sfax University. She received her Ph.D. in computer Science in 2016. She iscurrently a research member in the MIRACL (Multimedia InfoRmation system and Advanced Computing Laboratory) laboratory. Her Ph.D. work aims to integrate morpho-syntactic and semantic information for statistical machine translation. Her main interest focuses on Arabic language processing.

Salma Jamoussi, University of Sfax, MIRACL Laboratory

Researcher and assistant professor at Sfax University in the higher institute of computer science and multimedia. She received her Ph.D. in computer Science in 2004 from the Henri Poincaré University, France. She focuses her research on classification methods, datamining and natural language processing.

Abdelmajid Ben Hamadou, University of Sfax, MIRACL Laboratory

Obtained a doctorate degree in computer science from the University of Orsay (France) in November 1979 and a These d′Etat in Computer Science from the University of Tunis (Tunisia) in March 1993. He is presently Professor of Computer Science at the Higher Institute of Computer science and Multimedia, Sfax-university and member of the Research Laboratory MIRACL at the same institution. In July 2002 he was decorated by the President of the Tunisian Republic (“Merit in Education and Science”) and in May 2009, he received from the Vice President of Syria the "Al-Kindi" Award (“the best computer science researcher”). Abdelmajid Ben Hamadou has published more than 280 articles in journals and conferences and has supervized more than 40 doctoral theses. His research domains are: Natural Language Processing, semantic web, information retrieval/filtering and document summarizing.

Descargas

Publicado

2016-12-18

Número

Vol. 20 Núm. 4 (2016): Thematic Issue: Research Advances and Applications of Evolutionary Computation (Guest Editors: H.J. Escalante, M. Graff)

Sección

Artículos

Licencia

Transfiero exclusivamente a la revista “Computación y Sistemas”, editada por el Centro de Investigación en Computación (CIC), los Derechos de Autor del artículo antes mencionado, asimismo acepto que no serán transferidos a ninguna otra publicación, en cualquier formato, idioma, medio existente (incluyendo los electrónicos y multimedios) o por desarrollar.

Certifico que el artículo, no ha sido divulgado previamente o sometido simultáneamente a otra publicación y que no contiene materiales cuya publicación violaría los Derechos de Autor u otros derechos de propiedad de cualquier persona, empresa o institución. Certifico además que tengo autorización de la institución o empresa donde trabajo o estudio para publicar este Trabajo.

El autor, representante acepta la responsabilidad por la publicación del Trabajo en nombre de todos y cada uno de los autores.

Esta Transferencia está sujeta a las siguientes reservas:

  • Los autores conservan todos los derechos de propiedad (tales como derechos de patente) de este Trabajo, con excepción de los derechos de publicación transferidos al CIC, mediante este documento.
  • Los autores conservan el derecho de publicar el Trabajo total o parcialmente en cualquier libro del que ellos sean autores o editores y hacer uso personal de este trabajo en conferencias, cursos, páginas web personal, etc.