N-gram and Word2Vec Feature Engineering Approaches for Spam Recognition on Some Influential Twitter Topics in Saudi Arabia
Bibliographic record
Abstract
Social media platforms, such as Twitter, have become powerful sources of information on people's perception of major events. Many people use Twitter to express their views on various issues and events and use it to develop their opinion on the diverse economic, political, technical, and social occurrences related to their daily lives. Spam and non-relevant tweets are a major challenge for Twitter trend detection. Saudi Arabia is a top ranked country in Twitter usage worldwide, and in recent years has experienced difficulties due to the use and rise of hashtags based on misleading tweets and spam. The goal of this paper is to apply machine learning techniques to identify spam on the Saudi tweets collected to the end of 2020. To date, spam detection on Twitter data has been mostly done in English, leaving other major languages, such as Arabic, insufficiently covered. Additionally, publicly accessible Arabic Twitter datasets are hard to find. For our research, we use eight Twitter datasets on some significant topics in politics, health, national affairs, economy, and sport, to train and evaluate different machine learning algorithms, with a focus on two feature generation techniques based on N-grams and Word2Vec embeddings. One contribution of this paper is providing these new labelled datasets with embeddings. The experimental results show improvement from using embeddings over N-grams in more balanced datasets vs. more unbalanced ones. We also find a superior performance of the Random Forest algorithm over other algorithms in most experiments.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".