A Language Model for Arabic Texts Disambiguation using Deep Learning

Abdel Rahman, Nehad M.; Nouh, Sayed A.; Abo Al-ez, Reda H.

doi:10.21608/ejle.2019.59143

|
|
|
How to cite
RIS EndNote BibTeX APA MLA Harvard Vancouver
|
Share

	A Language Model for Arabic Texts Disambiguation using Deep Learning
Article 1, Volume 6, Issue 2, September 2019, Page 1-16 PDF (755.06 K)
Document Type: Original Article
DOI: 10.21608/ejle.2019.59143
View on SCiNiTO
Authors
Nehad M. Abdel Rahman ¹; Sayed A. Nouh²; Reda H. Abo Al-ez²
¹Computer science and Information Technology College at Imam Abdulrahman bin Faisal University, Dammam, Saudi Arabia
²Computers and Systems Engineering Department at Al-Azhar University, Cairo, Egypt
Abstract
Careful learning and understanding of Arabic content are extremely important because of its continuous growing. There are many types of text learning. It includes summarization, translation, and classification. The choice of the Arabic language in this research came from the lack of modern research such as deep learning. Although deep learning of Arabic is not a goal in this research, the main goal was to remove ambiguity in the Arabic language, since Arabic has more words than meaning, according to diacritics and the grammar of the text. The research idea is based on deep text learning, text analysis and removing of the ambiguity. In this paper, we proposed a new method for Arabic text learning by using deep learning methods. we use in this method the learning word vectors as weights by using 2000000-word vectors. The language model and the word analysis were also used to analyze the text and to detect the ambiguous words. Additionally, the learning results from the deep learning were compared with other researches of text from an accuracy perspective.
Keywords
Disambiguation; Deep learning; Convolution Neural Network; Long-Short Term Memory; natural language processing; Word Embeddings; Clustering; classification; and Language Model


Statistics Article View: 406 PDF Download: 892