Hate speech detection in Arabic social networks using deep learning and fine-tuned embeddings
Bibliographic record
Abstract
In recent years, opinions and communication can be easily expressed through social media networks that have allowed users to communicate and share their opinions and views, resulting in massive user-generated content. This content may contain text that is hateful to large groups or specific individuals. Therefore, in most website policies, automatic hate speech detection is required, and early automatic detection or filtering of such content is critical and necessary in online social networks, especially with large and increasingly user-generated content. This paper presents a suggested model to enhance the detection performance of hate speech using deep learning models with two types of word embedding models, the first model is Arabic models based on Wor2Vec including AraVec and Mazajak. The second is word embedding techniques models based on BERT including three pre-trained models namely ARABERT, MARBERT and CAMeLBERT. Common metrics in text classification are used including precision, recall, accuracy, and F1 score for model assessment. The experimental results show fine-tuned Arabic BERT models outperform Word2Vec based models, and that MARBERT outperforms both ARABERT and CAMeLBERT across all deep learning architectures, highlighting its superior ability to classify Arabic text. Additionally, BLSTM models show the highest performance on ARABERT, MARBERT, and CAMeLBERT, achieving an accuracy of 0.9945 with MARBERT.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".