Skip to main content
← Back to Projects

AI / NLP

Knowledge Graph Q&A System

An end-to-end pipeline that scrapes web data, extracts entities, builds a knowledge graph in Neo4j, and enables natural language question answering.

Problem

Unstructured web data is difficult to query and reason about. Traditional search fails to capture relationships between entities.

Solution

Built an automated pipeline: web scraping → entity extraction (NER) → relationship mapping → Neo4j storage → graph-based Q&A system.

Architecture

1

Web Scraping

Automated data collection from web sources

2

Entity Extraction

NER-based entity identification

3

Relationship Mapping

Entity relationship identification

4

Neo4j Storage

Graph database for structured storage

5

Q&A Interface

Natural language query interface

Technologies

PythonNeo4jNLPWeb ScrapingKnowledge GraphsNER

Lessons Learned

  • Graph databases excel at relationship-heavy queries

  • NER accuracy directly impacts downstream graph quality

  • Web scraping requires robust error handling and rate limiting