Enhancing Detection of Arabic Social Spam Using Data Augmentation and Machine Learning

Abstract

In recent years, people have tended to use online social platforms, such as Twitter and Facebook, to communicate with families and friends, read the latest news, and discuss social issues. As a result, spam content can easily spread across them. Spam detection is considered one of the important tasks in text analysis. Previous spam detection research focused on English content, with less attention to other languages, such as Arabic, where labeled data are often hard to obtain. In this paper, an integrated framework for Twitter spam detection is proposed to overcome this problem. This framework integrates data augmentation, natural language processing, and supervised machine learning algorithms to overcome the problems of detection of Arabic spam on the Twitter platform. The word embedding technique is employed to augment the data using pre-trained word embedding vectors. Different machine learning techniques were applied, such as SVM, Naive Bayes, and Logistic Regression for spam detection. To prove the effectiveness of this model, a real-life data set for Arabic tweets have been collected and labeled. The results show that an overall improvement in the use of data augmentation increased the macro F1 score from 58% to 89%, with an overall accuracy of 92%, which outperform the current state of the art.

Dataset of Arabic spam and ham tweets

“…So, the percentage of ham tweets is double the percentage of spam tweets. The dataset in [9] used only one single hashtag to collect all the data. So, the classifier will only fit this specific data.…”

“…While the available dataset is sufficient for training deep learning models, it is valuable to augment it to increase the diversity and quality of data. Data augmentation [9] , particularly, helps to balance the number of spam and ham tweets. However, augmenting Arabic text data in natural language processing presents challenges due to the complex nature of language.…”

Breast cancer diagnosis and management guided by data augmentation, utilizing an integrated framework of SHAP and random augmentation

“…Several approaches have been suggested to address the aforementioned issues with the use of ML, with data augmentation being one of the most popular. [8][9][10][11][12] Data augmentation involves modifying or adding features to the original data to generate new data. It is often employed to minimize significant errors that may arise when building diagnostic models using small datasets.…”

“…Several approaches have been suggested to address the aforementioned issues with the use of ML, with data augmentation being one of the most popular 8–12 . Data augmentation involves modifying or adding features to the original data to generate new data.…”

KnowTellConvince at ArAIEval Shared Task: Disinformation and Persuasion Detection in Arabic using Similar and Contrastive Representation Alignment

“…Rumors, highly contagious within tight-knit Arabic communities, necessitate vigilant monitoring to counteract panic and misinformation, exploiting cultural contexts for added complexity Harrag and Djahli, 2022). Spam, spanning fraudulent ads and misleading claims, pervades digital spaces in all languages, underlining the need to distinguish it from credible content for online source credibility (Kaddoura et al, 2023;Alkadri et al, 2022). Propaganda, a pivotal element of disinformation campaigns, influences public opinion and necessitates understanding and countering within the Arabic-speaking context to protect individuals and communities from manipulation by misleading narratives (Sharara et al, 2022;Feldman et al, 2021).…”