Al-Farahidi Arabic diacritizer

Abstract EN

The paper proposes an automatic Arabic Diacritizer namely, AL-Farahidi Arabic Diacritizer, AFAD.

The proposed AFAD is based on a hybrid approach that combines the Statistical as well as the Morphological methods.

The validation of AL-Farahidi Arabic Diacritizer is conducted making use of generated tens of sentences.

Three metrics were used to check its performed adequacy.

The achieved performance of AFAD include 92% Word Error Rate, 90% Diacritic Error Rate and 80% Sentence Error Rate.

A performance comparison is accomplished between AFAD and the other available diacritizers namely, Mishkal, RDI and the MADAMIA.

The performance of AL-Farahidi Arabic Diacritizer outperforms all other three diacritizers.

Although, we have achieved these preliminary and promising results, it is still too early to declare its overwhelmed performance.

As it requires further investigation and expansion such as testing the speed of processing, increasing the number of regular and irregular grammatical sentences, longer sentences, which will be the future work.