Sovereign 8s.is mirror of standardgalactic/Amharic-News-Classification
  • Jupyter Notebook 98.6%
  • Python 1.4%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2026-01-09 17:05:45 +03:00
amharic-news-ontology-ontologies.owx.zip Add files via upload 2025-11-28 21:17:15 +03:00
Annotation Guidelines and Protocols Add methodology for Amharic news ontology construction 2026-01-09 11:36:51 +03:00
comparassion.ipynb Add files via upload 2026-01-09 17:05:45 +03:00
m4.py Add files via upload 2025-11-28 21:16:29 +03:00
main_onto2.py Add files via upload 2025-11-28 21:16:07 +03:00
README.md Update README.md 2025-11-28 21:13:09 +03:00

Amharic-News-Classification

Amharic news classification using ontology and machine learning"

Amharic News Ontology (ANO)

A comprehensive OWL ontology for Amharic news classification and semantic analysis. This ontology provides structured knowledge representation for Ethiopian news domains.

📁 Files

  • amharic-news-ontology.owl - Main ontology file in OWL/XML format

🏗️ Ontology Structure

Core Classes

  • NewsArticle - Base class for news articles
  • NewsCategory - Main news categories with subclasses:
    • Politics (ፖለቲካ)
    • Sports (ስፖርት)
    • Business (ቢዝነስ)
    • Entertainment (መዝናኛ)
    • NationalNews (ሀገር አቀፍ ዜና)
    • InternationalNews (አለም አቀፍ ዜና)

Subclass Hierarchy

Each main category has specialized subclasses:

  • Politics: Elections, Governance, ForeignPolicy, Legislation
  • Sports: Football, Athletics, Basketball, Olympics
  • Business: Economy, Finance, Trade, Market
  • Entertainment: Music, Movies, Theater, Arts

Object Properties

  • hasTopic - Links articles to categories
  • isRelatedTo - Defines semantic relationships
  • isPartOf - Represents hierarchical composition
  • hasSubEvent - Captures event relationships

Data Properties

  • hasFrequency - Concept occurrence tracking
  • hasDiscriminativePower - Feature importance scoring
  • lastUpdated - Ontology maintenance timestamp

Individuals

187 domain-specific instances including:

  • Political parties (Prosperity Party, Ezema Party)
  • Sports teams (Ethiopia National Team, St. George FC)
  • Business entities (Commercial Bank, Ethio Telecom)
  • Entertainment personalities (Teddy Afro)
  • Geographic locations (Addis Ababa, United Nations)

🔧 Tools for Editing

  1. Protégé (Professional) - Download from protege.stanford.edu
  2. WebProtégé (Web-based) - Access at webprotege.stanford.edu
  3. VS Code with OWL extensions

VS Code Extensions:

# Install these extensions:
- OWL Syntax Highlighting
- XML Tools
- RDF and OWL

# Amharic News Classification

A machine learning system that classifies Amharic news articles using ontology-based feature fusion and logistic regression.

## 🚀 Quick Start

```bash
# Install dependencies
pip install -r requirements.txt

# Train model
python main.py

# Use for prediction
from classifier import AmharicNewsClassifier
model = AmharicNewsClassifier()
model.load_model("amharic_news_classifier.pkl")
result = model.predict("Your Amharic text here")
src/
├── main.py              # Main training script
├── classifier.py        # Model class
├── preprocessor.py      # Text preprocessing
└── ontology_processor.py # Semantic features