Comparison between two Arabic tagsets

Author

Rashwan, Mohsen A A ; Khalil, Enas A H ; Rafea, Ahmed

Author_Institution

Dept. of Electron. & Electr. Commun., Cairo Univ, Cairo, Egypt

fYear

2009

fDate

24-27 Sept. 2009

Firstpage

1

Lastpage

8

Abstract

Enhancing Arabic tagging is of great importance in many NLP applications. This paper presents a simple comparison tool that compares two powerful tagging systems for Arabic, the first one is the ASVM Tagger, by Diab M. et al,. The second one is RDI Arab Tagger that relies on simple powerful long n-grams probability estimation plus A*search algorithm for disambiguation, this comparison is done to superimpose points of excellence in Arab Tagger into ASVM tagger. From this comparison, mapper tool is implemented to convert from the fine grain Arab tagset (62 tags used by the ArabTagger) to the other course grain compact tagset of 24 tags Reduced Tagset (RTS) used by ASVM-Tagger. A combined system from the output of both is then formed, which gives an average accuracy higher than that of ASVM in our experiment, 95% of hybrid system versus 93% of ASVM system.

Keywords

natural language processing; probability; support vector machines; tree searching; A* search algorithm; ASVM Tagger; ArabTagger; Arabic tagging; NLP applications; RDI Arab Tagger; long n-grams probability estimation; natural language processing; reduced tagset; Application software; Computer science; Data mining; Labeling; Machine learning; Natural languages; Power engineering and energy; Speech processing; Support vector machines; Tagging; A∗search algorithm; Automatic Support Vector Machine (ASVM); N-gram model; Part-of-Speech Tagging (POS); Reduced Tag Set (RTS);

fLanguage

English

Publisher

ieee

Conference_Titel

Natural Language Processing and Knowledge Engineering, 2009. NLP-KE 2009. International Conference on

Conference_Location

Dalian

Print_ISBN

978-1-4244-4538-7

Electronic_ISBN

978-1-4244-4540-0

Type

conf

DOI

10.1109/NLPKE.2009.5313767

Filename

5313767