AH
Abas Hussein Ali
Research and development team member
Analyze Somali political claims with eight machine-learning models, inspect source evidence, and submit important cases for human fact-checking.
PMD separates the model prediction from supporting evidence and the final decision of a fact-checker.
Traditional, transformer, and deep-learning classifiers predict Reliable, Misinformation, or Misleading.
All available models can vote on the same text while model agreement remains separate from prediction confidence.
Related web and social sources are collected for review as supporting, contradicting, repeating, or unclear evidence.
Cases can be assigned to fact-checkers, who publish a separate decision, explanation, and evidence record.
The system moves from automated analysis to evidence collection and an optional human decision.
Paste a Somali claim, news excerpt, or social-media post.
The system checks Somali language and whether the content concerns politics or public affairs.
Choose one model or use the ensemble to compare all available classifiers.
Review similar reports, trusted-source matches, dates, and source links.
A fact-checker can confirm a label or mark the evidence as insufficient.
Eight available classifiers can be used individually or together through the ensemble.
A Somali transformer classifier fine-tuned for the three project labels.
A second Somali transformer classifier used for individual and ensemble predictions.
A support vector classifier using TF-IDF features for Somali text.
A linear multiclass classifier using TF-IDF text features.
A tree-based classifier included as an individual ensemble voter.
A probabilistic text classifier included as an individual ensemble voter.
Combines convolutional feature extraction with bidirectional sequence learning for Somali text classification.
Processes Somali text in both directions to capture contextual relationships between words.
Try up to three Somali political claims. The ensemble is selected by default.
We are a final-year computing team at JUST University developing PMD to support Somali political-text analysis, transparent source review, and responsible human fact-checking.
Research and development team member
Research and development team member
Research and development team member
Research and development team member
Contact the PMD team about research collaboration, technical questions, fact-checker participation, or responsible use of the system.