Word Representation Models for Arabic Dialect Identification

International Journal of Advanced Computer Science and Applications, 2022

Author profiling is the process of finding characteristics that make up an author's profile. This paper presents a machine learning-based author profiling model for Arabic users, considering the author's regional dialect as a crucial characteristic. Various classification algorithms have been implemented: decision tree, KNN, multilayer perceptron, random forest, and support vector machines. A pair of Continuous Bag-of-Word (CBOW) models has been used for word representation. A well-known data set has been used to evaluate the proposed model and a data augmentation process has been implemented to improve the quality of training data. Support vector machines achieved a 50.52% f1-score, outperforming other models.

2020

We present the Arabic dialect identification system that we used for the country-level subtask of the NADI challenge. Our model consists of three components: BiLSTM-CNN, character-level TF-IDF, and topic modeling features. We represent each tweet using these features and feed them into a deep neural network. We then add an effective heuristic that improves the overall performance. We achieved an F1-Macro score of 20.77% and an accuracy of 34.32% on the test set. The model was also evaluated on the Arabic Online Commentary dataset, achieving results better than the state-of-the-art.

2020

This paper describes Faheem (adj. of understand), our submission to NADI (Nuanced Arabic Dialect Identification) shared task. With so many Arabic dialects being understudied due to the scarcity of the resources, the objective is to identify the Arabic dialect used in the tweet, at the country-level. We propose a machine learning approach where we utilize word-level ngram (n = 1 to 3) and tf-idf features and feed them to six different classifiers. We train the system using a data set of 21,000 tweets—provided by the organizers—covering twenty-one Arab countries. Our top performing classifiers are: Logistic Regression, Support Vector Machines, and Multinomial Naı̈ve Bayes (MNB). We achieved our best result of macro-F1 = 0.151 using the MNB classifier.

2018

This paper describes our Automatic Dialect Recognition (ADI) system for the VarDial 2018 challenge, with the goal of distinguishing four major Arabic dialects, as well as Modern Standard Arabic (MSA). The training and development ADI VarDial 2018 data consists of 16,157 utterances, their words transcription, their phonetic transcriptions obtained with four non-Arabic phoneme recognizers and acoustic embedding data. Our overall system is a combination of four different systems. One system uses the words transcriptions and tries to recognize the speaker dialect by modeling the sequence of words for each dialect. Another system tries to recognize the dialect by modeling the phones sequence produced by non-Arabic phone recognizers, whereas, the other two systems use GMM trained on the acoustic features for recognizing the dialect. The best performance was achieved by the fused system which combines four systems together, with F1 micro of 68.77%.

2020

Arabic dialect identification is a complex problem for a number of inherent properties of the language itself. In this paper, we present the experiments conducted, and the models developed by our competing team, Mawdoo3 AI, along the way to achieving our winning solution to subtask 1 of the Nuanced Arabic Dialect Identification (NADI) shared task. The dialect identification subtask provides 21,000 country-level labeled tweets covering all 21 Arab countries. An unlabeled corpus of 10M tweets from the same domain is also presented by the competition organizers for optional use. Our winning solution itself came in the form of an ensemble of different training iterations of our pre-trained BERT model, which achieved a micro-averaged F1-score of 26.78% on the subtask at hand. We publicly release the pre-trained language model component of our winning solution under the name of Multi-dialect-Arabic-BERT model, for any interested researcher out there.

Dialect identification is a prior requirement for learning lexical and morphological knowledge a language variation that can be beneficial for natural language processing (NLP) and potential AI downstream tasks. In this paper, we present the first work on sentence-level Modern Standard Arabic (MSA) and Saudi Dialect (SD) identification where we trained and tested three classifiers (Logistic regression, Multi-nominal Na¨ıve Bayes, and Support Vector Machine) on datasets collected from Saudi Twitter and automatically labeled as (MSA) or SD. The model for each configuration was built using two levels of language models, i.e., unigram and bi-gram, as feature sets for training the systems. The model reported high-accuracy performance using 10-fold cross-validations with average 98.98%. This model was evaluated on another unseen, manually-annotated dataset. The best performance of these classifiers was achieved by Multi-nominal Naïve Bayes, reporting 89%.

Proceedings of the Fourth Arabic Natural Language Processing Workshop, 2019

Studies on Dialectical Arabic are growing more important by the day as it becomes the primary written and spoken form of Arabic online in informal settings. Among the important problems that should be explored is that of dialect identification. This paper reports different techniques that can be applied towards such goal and reports their performance on the Multi Arabic Dialect Applications and Resources (MADAR) Arabic Dialect Corpora. Our results show that improving on traditional systems using frequency based features and non deep learning classifiers is a challenging task. We propose different models based on different word and document representations. Our top model is able to achieve an F1 macro averaged score of 65.66 on MADAR's smallscale parallel corpus of 25 dialects and Modern Standard Arabic (MSA).

2020

In this paper, we present three methods developed by the SORBONNE Team for the NADI shared task on Arabic Dialect Identification for tweets. The first and the second method use respectively a machine learning model based on a Voting Classifier with words and character level features and a deep learning model at the word level. The third method uses only character-level features. We explored different text representation such as TF-IDF (first model) and word embeddings (second model). The Voting Classifier was the most powerful prediction model, achieving the best macro-average F1 score of 18.8% and an accuracy of 36.54% on the official test. Our model ranked 9 on the challenge and in conclusion we propose some ideas to improve its results.

Communications in Computer and Information Science, 2016

We present a study on sentence-level Arabic Dialect Identification using the newly developed Multidialectal Parallel Corpus of Arabic (MPCA) -the first experiments on such data. Using a set of surface features based on characters and words, we conduct three experiments with a linear Support Vector Machine classifier and a meta-classifier using stacked generalization -a method not previously applied for this task. We first conduct a 6-way multi-dialect classification task in the first experiment, achieving 74% accuracy against a random baseline of 16.7% and demonstrating that meta-classifiers can large performance increases over single classifiers. The second experiment investigates pairwise binary dialect classification within the corpus, yielding results as high as 94%, but also highlighting poorer results between closely related dialects such as Palestinian and Jordanian (76%). Our final experiment conducts cross-corpus evaluation on the widely used Arabic Online Commentary (AOC) dataset and demonstrates that despite differing greatly in size and content, models trained with the MPCA generalize to the AOC, and vice versa. Using only 2,000 sentences from the MPCA, we classify over 26k sentences from the radically different AOC dataset with 74% accuracy. We also use this data to classify a new dataset of MSA and Egyptian Arabic tweets with 97% accuracy. We find that character n-grams are a very informative feature for this task, in both within-and cross-corpus settings. Contrary to previous results, they outperform word n-grams in several experiments here. Several directions for future work are outlined.

2021

In this paper, we analyze the impact of the weighted concatenation of TF-IDF features for the Arabic Dialect Identification task while we participated in the NADI2021 shared task. This study is performed for two subtasks: subtask 1.1 (country-level MSA) and subtask 1.2 (country-level DA) identification. The classifiers supporting our comparative study are Linear Support Vector Classification (LSVC), Linear Regression (LR), Perceptron, Stochastic Gradient Descent (SGD), Passive Aggressive (PA), Complement Naive Bayes (CNB), MutliLayer Perceptron (MLP), and RidgeClassifier. In the evaluation phase, our system gives F1 scores of 14.87% and 21.49%, for country-level MSA and DA identification respectively, which is very close to the average F1 scores achieved by the submitted systems and recorded for both subtasks (18.70% and 24.23%).