📖 Story2KG: From Narratives to Knowledge Graphs

Transform unstructured stories into structured, queryable Knowledge Graphs using cutting-edge NLP.

---

🌟 Overview

Story2KG is an end-to-end Natural Language Processing (NLP) pipeline that transforms raw narrative text into structured Knowledge Graphs (KGs).

Unlike prior works focusing on individual subtasks (NER, SRL, summarization), Story2KG provides a unified framework for:

• 🧹 Preprocessing & Coreference Resolution
• 📚 Scene Segmentation
• 🔎 Entity, Attribute, and Event Extraction
• 😃 Emotion Classification
• ✍️ Abstractive Summarization (BART/T5)
• 🌐 Knowledge Graph Construction in Neo4j

The result is an interactive, queryable graph that captures characters, events, emotions, and relationships from stories.

---

🚀 Key Features

✅ End-to-End Pipeline – From raw text to KG.
✅ Narrative-Aware Design – Handles stories, characters, events, and temporal flow.
✅ Hybrid NLP Models – SpaCy, Hugging Face Transformers, AllenNLP.
✅ Graph Visualization – Neo4j integration for querying and visualization.
✅ Modular Architecture – Swap models (e.g., BART → T5).
✅ Applications – Education, Digital Humanities, Explainable AI, Story Analytics.

---

🏗️ Pipeline Architecture

graph TD
    A[Raw Story Text] --> B[Preprocessing & Coreference Resolution]
    B --> C[Scene Segmentation]
    C --> D[Deep Context Extraction: Entities, Attributes, Events, Emotions]
    D --> E[Abstractive Summarization]
    E --> F[Neo4j Knowledge Graph Construction]

Loading

---

📂 Project Structure

Story2KG/
│── notebooks            # Jupyter notebooks for experimentation
│── Architectures        # Adaptive + Hierarchical
│── requirements.txt     # Dependencies
│── License              # MIT License
│── README.md            # Project Documentation

---

⚙️ Installation

1. Clone the Repository

git clone https://github.com/yourusername/Story2KG.git
cd Story2KG

2. Create Virtual Environment

python -m venv venv
source venv/bin/activate   # (Linux/Mac)
venv\Scripts\activate      # (Windows)

3. Install Dependencies

pip install -r requirements.txt

4. Setup Neo4j

• Install Neo4j Desktop or run a Docker container:

docker run -d --name neo4j -p 7474:7474 -p 7687:7687 -e NEO4J_AUTH=neo4j/test neo4j:latest

---

🖥️ Usage

Run Pipeline on a Story

from story2kg import pipeline

story_text = """Once, a hare laughed at a tortoise for being slow..."""
kg = pipeline.run(story_text)

# Export to Neo4j
kg.export_to_neo4j(uri="bolt://localhost:7687", user="neo4j", password="test")

Query Example (Cypher)

MATCH (c:Character)-[r]->(e:Event)
RETURN c.name, type(r), e.description;

---

📊 Evaluation

Component	Precision	Recall	F1-Score
Entity Recognition	89.0%	91.2%	90.1%
Attribute Detection	85.4%	83.7%	84.5%
Emotion Classification	80.2%	78.9%	79.5%
Knowledge Graph Completeness	–	–	92.5%

---

🔮 Future Work

• Expansion to larger story corpora (novels, folklore).
• Integration with LLMs (GPT, LLaMA) for improved reasoning.
• Support for Temporal & Causal KGs.
• Advanced narrative coherence evaluation metrics.

---

🎯 Applications

📘 Education – Transform textbooks/fables into interactive graphs.
📖 Digital Humanities – Analyze cultural narratives & folklore.
🤖 Explainable AI – Human-readable narrative reasoning.
📝 Story Analytics – Character profiling & plot analysis.

---

👨‍💻 Contributors

• Dr. Mohan Allam – Advisor
• Dasara Rajiv Kumar – Co-Author
• Rohit Mukkala – Co-Author
• Sujal Ghonmode – Co-Author

---

📜 License

This project is licensed under the MIT License. See LICENSE for details.