Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
February 2, 2026PeerJ Computer ScienceOpen Access

An Amharic sexually explicit content detection model using fine-tuned bidirectional encoder representation from transformers and explainable artificial intelligence

View Full Paper
Ask AI
Bookmark
Share

Authors

DEDemeke Endalie

Discussion

Loading...

Member takes

Implication

This model detects sexually explicit content in Amharic, suggesting a reliable tool for digital safety measures.

Key Points

  • To develop a model for detecting sexually explicit content in the Amharic language.
  • Utilized a fine-tuned BERT model trained on previously annotated hate speech data.
  • Annotated a dataset of 34,710 comments collected from various social media platforms.
  • Applied Explainable AI techniques, specifically LIME, to interpret model decisions.
  • Achieved accuracy of 94%, precision of 95%, recall of 94%, and F1-score of 94%.
  • Revealed how the model classifies content by computing probabilities and highlighting significant words.
  • Outperformed multiple state-of-the-art text classification methods, including BERT, SVM, MLP, and CNN.

Cite This Study

Demeke Endalie (2026) studied this question.

synapsesocial.com/papers/6980fe57c1c9540dea8105afhttps://doi.org/10.7717/peerj-cs.3529
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Domain-specific language models pre-trained on construction management systems corpora2024 · 63 citations
  2. 2Text Independent Amharic Language Dialect Recognition: A Hybrid Approach of VQ and GM2017 · 6 citations
  3. 3The Detection of Sexual Harassment and Chat Predators Using Artificial Neural Network2021 · 18 citations
  4. 4Socio-cultural factors perceived to influence sexual behaviours of adolescents in Ethiopia2023 · 6 citations
  5. 5Amharic Adhoc Information Retrieval System Based on Morphological Features2022 · 9 citations