DreamRunnerMoshi/Neo4JLLM

Bangla Archive Data Graph Knowledge Base # Bengali Newspaper Knowledge Graph with Named Entity Recognition A comprehensive project demonstrating the integration of Natural Language Processing (NLP) and Graph Database technologies for analyzing Bengali newspaper articles. This project combines data extraction, Named Entity Recognition (NER), and graph database modeling to create a structured knowledge representation of Bengali news content. ## 🎯 Project Overview This project showcases the complete pipeline for: - **Data Acquisition**: Downloading and processing Bengali newspaper datasets from Kaggle - **Data Modeling**: Structuring news articles in a Neo4j graph database - **Named Entity Recognition**: Extracting entities from Bengali text using state-of-the-art transformer models - **Knowledge Graph Construction**: Building relationships between newspapers, articles, and extracted entities ## 🔧 Technologies Used - **Python 3.x**: Core programming language - **Neo4j**: Graph database for storing and querying knowledge graphs - **Transformers**: Hugging Face library for NLP models - **Bengali NER Model**: `Suchandra/bengali_language_NER` - Specialized model for Bengali text - **Jupyter Notebook**: Interactive development environment - **JSON**: Data format for news articles ## 📊 Dataset - **Source**: Bangla Newspaper Dataset from Kaggle - **Format**: JSON containing structured news articles - **Content**: Bengali newspaper articles with metadata including: - Newspaper name - Headlines - Article body - Publication date - Categories and subcategories - Source links ## 🏗️ Architecture ``` Data Pipeline Flow: Kaggle Dataset → JSON Processing → Neo4j Graph Database → NER Analysis → Entity Extraction ``` ### Graph Database Schema The Neo4j database uses the following node and relationship structure: - **Nodes**: - `Newspaper`: Represents news publications - `Article`: Individual news articles with full content and metadata - `Entity`: Extracted named entities (Person, Location, Organization, etc.) - * …