Social Media Hate Speech Detection Using Explainable Artificial Intelligence (XAI)
Bibliographic record
Abstract
Explainable artificial intelligence (XAI) characteristics have flexible and multifaceted potential in hate speech detection by deep learning models. Interpreting and explaining decisions made by complex artificial intelligence (AI) models to understand the decision-making process of these model were the aims of this research. As a part of this research study, two datasets were taken to demonstrate hate speech detection using XAI. Data preprocessing was performed to clean data of any inconsistencies, clean the text of the tweets, tokenize and lemmatize the text, etc. Categorical variables were also simplified in order to generate a clean dataset for training purposes. Exploratory data analysis was performed on the datasets to uncover various patterns and insights. Various pre-existing models were applied to the Google Jigsaw dataset such as decision trees, k-nearest neighbors, multinomial naïve Bayes, random forest, logistic regression, and long short-term memory (LSTM), among which LSTM achieved an accuracy of 97.6%. Explainable methods such as LIME (local interpretable model—agnostic explanations) were applied to the HateXplain dataset. Variants of BERT (bidirectional encoder representations from transformers) model such as BERT + ANN (artificial neural network) with an accuracy of 93.55% and BERT + MLP (multilayer perceptron) with an accuracy of 93.67% were created to achieve a good performance in terms of explainability using the ERASER (evaluating rationales and simple English reasoning) benchmark.
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 | 0.002 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".