A Stochastic Arabic Diacritizer Based on a Hybrid of Factorized and Unfactorized Textual Features

Rashwan, Mohsen; Al-Badrashiny, Mohamed; Attia, Mohamed; Abdou, Sherif M.; Rafea, Ahmed

doi:10.1109/tasl.2010.2045240

Cited by 52 publications

(45 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, on ATB standard dataset, the proposed system achieves 0.9% improvement over the best result in literature using the same training and testing data same as evaluation in Rashwan et al (2011) was done.…”

Section: Comparison To Other Systemsmentioning

confidence: 84%

“…There are many models for Arabic PoS tags. In this work we adopt the one in Rashwan et al (2011), which sets 62 context-free atomic units to represent all possible Arabic language PoS tags. A very rich dataset of Arabic words, extracted from different sources, is used to train the system (available on http://www.RDIeg.com/RDI/TrainingData is where to download TRN_DB_II).…”

Section: Context Class Labelsmentioning

confidence: 99%

“…In addition, the diacritics classification can be divided into syntactical diacritization, caring about caseending and morphological diacritization, caring about the rest of the word diacritics. So far, morphological part of the problem is almost solved, leaving a marginal error of around 3-4%, Rashwan et al (2009Rashwan et al ( , 2011. On the other hand, syntactical diacritization errors are still high, hitting a ceiling that is claimed to be asymptotic and cannot be squeezed any further, Rashwan et al (2009Rashwan et al ( , 2011.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automatic Arabic diacritics restoration based on deep nets

Sallab

Rashwan

Raafat

et al. 2014

Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP)

Self Cite

View full text Add to dashboard Cite

In this paper, Arabic diacritics restoration problem is tackled under the deep learning framework presenting Confused Subset Resolution (CSR) method to improve the classification accuracy, in addition to Arabic Part-of-Speech (PoS) tagging framework using deep neural nets. Special focus is given to syntactic diacritization, which still suffer low accuracy as indicated by related works. Evaluation is done versus state-of-the-art systems reported in literature, with quite challenging datasets, collected from different domains. Standard datasets like LDC Arabic Tree Bank is used in addition to custom ones available online for results replication. Results show significant improvement of the proposed techniques over other approaches, reducing the syntactic classification error to 9.9% and morphological classification error to 3% compared to 12.7% and 3.8% of the best reported results in literature, improving the error by 22% over the best reported systems

show abstract

Section: Comparison To Other Systemsmentioning

confidence: 84%

Section: Context Class Labelsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automatic Arabic diacritics restoration based on deep nets

Sallab

Rashwan

Raafat

et al. 2014

Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Roth et al [31] enriched this analysis with the use of lemmas. Rashwan et al [32] developed an Arab stochastic diacritizer based on a hybrid approach of factored and not factored textual features. They introduced the stochastic system with dual methods to automatically diacritize plain Arabic text.…”

Section: Related Workmentioning

confidence: 99%

A Hybrid Approach for the Morpho-Lexical Disambiguation of Arabic

2016

J Inf Process Syst

View full text Add to dashboard Cite

In order to considerably reduce the ambiguity rate, we propose in this article a disambiguation approach that is based on the selection of the right diacritics at different analysis levels. This hybrid approach combines a linguistic approach with a multi-criteria decision one and could be considered as an alternative choice to solve the morpho-lexical ambiguity problem regardless of the diacritics rate of the processed text. As to its evaluation, we tried the disambiguation on the online Alkhalil morphological analyzer (the proposed approach can be used on any morphological analyzer of the Arabic language) and obtained encouraging results with an F-measure of more than 80%.

show abstract

“…In [15], an approach was proposed which combines lexical retrieval, bigram-based and SVM-statistical prioritized techniques. In [14], the authors proposed two methods: the first uses an n-gram statistical language model along with A * lattice search while the second method attempts to segment each Arabic word into all its possible morphological constituents then proceed in a similar way as the first one. The authors reported that their second approach gives better results.…”

Section: Related Workmentioning

confidence: 99%