A Hybrid Semantic Enrichment Approach for Multi-Label Toxic Speech Detection
Bibliographic record
Abstract
The rapid growth of digital communication has facilitated the spread of toxic speech, which can harm individuals or communities and often appears across multiple nuanced categories.These categories are difficult to detect in short texts due to semantic ambiguity, limited context, and label dependencies.This study introduces a Hybrid Semantic Enrichment with Convolutional Neural Network (HSE-CNN) approach to enhance multilabel toxic speech classification.The HSE-CNN model leverages semantic enrichment techniques such as back translation, text expansion, word sense disambiguation (WSD), and semantic similarity mapping to enrich the contextual meaning of input texts.Using an Indonesian social media dataset containing 13,169 entries labeled with 12 toxic speech categories, we conducted a series of experiments involving preprocessing, semantic enrichment, and classification using various deep learning models.The optimal configuration includes a learning rate of 0.001, batch size of 16, and training for 30 epochs.Our proposed model achieved an F1-score of 80%, accuracy of 93%, and AUC of 91%, demonstrating its superiority over non-enriched models.Compared to baseline models such as BiLSTM and BiGRU, the HSE-CNN yields a 6.7% improvement in accuracy and a 4.5% improvement in F1-score.These findings suggest that HSE-CNN offers a promising solution for toxic speech detection systems, especially in resourcelimited languages, with potential applications in digital content moderation, online safety initiatives, and public awareness enhancement.
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".