SNAD Arabic Dataset for Deep Learning

Alsaleh, Deem; Alamir, Mashael Bin; Marie-Sainte, Souad Larabi

doi:10.1007/978-3-030-55180-3_47

Cited by 3 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the same task, Almuzaini and Azmi [23] proposed seven DL models incorporating different stemming and embedding techniques to assess their impact on Arabic doc-ument classification. Two datasets were employed in their experiment: the Arabic News Texts (ANT) [24] and the Saudi Press Agency (SPA) [25]. The ANT dataset contained 10,161 documents across nine categories, while the SPA dataset included 13,402 documents categorized into six groups.…”

Section: A Multi-class Text Classificationmentioning

confidence: 99%

Exploring Character Trigrams for Robust Arabic Text Classification: A Comparative Analysis in the Face of Vocabulary Expansion and Misspelled Words

Alomari,

Ahmad

2024

IEEE Access

View full text Add to dashboard Cite

Tokenization is an important early step in natural language processing (NLP) tasks. The idea is to split the input sentence into smaller units, called tokens, for further processing. Words are the most commonly used tokens in text classification tasks but other tokenization ideas are also popular such as subword and character tokens. The increasing availability of training corpora has posed challenges for the word tokenization technique, primarily due to vocabulary size expansion. This has underscored the importance of exploring alternative tokenization, especially for morphologically rich languages like Arabic. In this study, we assess the efficacy of character trigrams for Arabic sentiment analysis and text classification, particularly their robustness against misspelled words. We compare character trigrams with word and WordPiece tokenization across five datasets, encompassing tweets sentiment analysis, reviews sentiment analysis, and news classification. These datasets, which range from small to large in vocabulary size, facilitate a comparative examination of vocabulary and Out-Of-Vocabulary (OOV) sizes for word and character trigram embeddings. The word and character trigram embeddings are integrated into a deep learning (DL) model featuring a dense layer connected to a Bidirectional Gated Recurrent Unit (BiGRU) for classification purposes. Meanwhile, WordPiece embeddings serve to fine-tune the AraBert model. Our findings reveal that the word embedding approach, when applied to extensive corpora, leads to a significant increase in vocabulary and OOV rates, whereas character trigram embedding maintains manageable sizes for both. On test sets devoid of spelling mistakes, the AraBert model surpasses the other models in performance. However, both word and character trigram embedding models exhibit similar levels of performance. In contrast, datasets with misspellings reveal a performance degradation for the WordPiece and word embedding models, with decline rates ranging between 2%-14% and 5%-19%, respectively. Meanwhile, the character trigram models exhibit stable performances, with a drop rate of 0%-8%. Notably, the performance of WordPiece and word embedding models suffers due to their inability to recognize misspelled words.

show abstract

Section: A Multi-class Text Classificationmentioning

confidence: 99%

Exploring Character Trigrams for Robust Arabic Text Classification: A Comparative Analysis in the Face of Vocabulary Expansion and Misspelled Words

Alomari,

Ahmad

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Table2presents the classes and their totals. The dataset can be downloaded[18] Details of Saudi Newspapers Articles Dataset…”

mentioning

confidence: 99%

“…DATA COLLECTIONIn this article, two datasets are used to experiment the proposed methodology. The first dataset was recently collected[18] from two major news sources in Saudi Arabia, AlRiyadh Newspaper and Saudi Press Agency (SPA). It consists of news text details and news titles classified into six classes: Economical, Political, Social, Sports,…”

mentioning

confidence: 99%

Arabic Text Classification Using Convolutional Neural Network and Genetic Algorithms

Alsaleh

Marie-Sainte

2021

IEEE Access

Self Cite

View full text Add to dashboard Cite

Arabic documents are massively rising due to numerous contents utilized in websites, social media, and news articles. The classification of such documents in labelled categories is a significant and vital task that deserves more attention. Arabic Text Classification is an emerging research theme in Arabic Natural Language Processing. Recently, Deep Neural Network approaches have successfully been applied to many text classification problems, especially in English Text Classification. Convolutional Neural Network (CNN) is one of the best popular models. However, CNN is not highly applied in Arabic Text Classification. In addition, the recent studies did not achieve a high classification accuracy due to parameter setting issue. To overcome this limitation, a new hybrid classification model for Arabic Text is developed. This paper proposes Genetic Algorithms based Convolutional Neural Network for Arabic Text Classification. Genetic Algorithm is used to optimize the CNN parameters. The proposed model is tested using two large datasets and compared with the state-of-the art studies. The results showed that the classification accuracy achieved an improvement of 4 to 5%.

show abstract