Agent Name CineGraph AI
What it does: CineGraph AI is a cinematic intelligence agent that transforms 100,000 movies into a reasoning engine. It performs multi-hop graph traversals to explain why certain creators and genres dominate the industry, moving beyond simple keyword search to true relational analysis.
Dataset and why a graph fits: I used the Global Movies Dataset (1950–2026), which includes 100,000 movies and over 400,000 relationships. Movies are inherently connected by Directors, Actors, Genres, and Eras. A graph is the perfect fit because it captures the causality chains—the ripple effect of a successful collaboration or a genre trend—that a flat table simply cannot represent.
- Scale: 115,000+ nodes, 400,000+ relationships.
Technology Stack
- Database: Neo4j Aura (Knowledge Graph)
- Reasoning Engine: Neo4j Aura Agent (Powered by LLM with multi-hop graph traversal)
- Data Pipeline: Python with
neo4jdriver andpandas - Visualization: Streamlit + Plotly (Live dashboard)
- Query Language: Cypher
Agent Tools The agent uses a combination of precise Cypher templates and natural language discovery:
- director_career: Analyzes entire filmographies, revenue tracking, and ratings.
- actor_collaborations: Evaluates the success of frequent actor-director pairings.
- genre_decade_trend: Tracks genre popularity and quality shifts from the 1950s to today.
- director_genre_country: Identifies a director's stylistic and geographic focus.
- blockbuster_formula: Pattern recognition for high-ROI movies.
- text2cypher: Fallback for ad-hoc natural language questions.
Example Conversation User: "How does Christopher Nolan's ROI in Sci-Fi compare to his work in other genres?" Agent Reasoning (Multi-Hop):
- Matches
(Director {name: 'Christopher Nolan'})-[:DIRECTED]->(m:Movie)— finds movies. - Follows
(m)-[:IN_GENRE]->(g:Genre)— separates filmography into Sci-Fi vs other genres. - Calculates
avg(m.roi_pct)for both groups — discovers Sci-Fi ROI is ~175%. - Synthesizes the answer, noting that his Sci-Fi work significantly outperforms his Dramas in profitability.
What makes this different
- Massive Scale: Processing 100K movies makes this one of the most comprehensive cinematic graphs in the competition.
- Specialized Tooling: Uses 6 custom Cypher templates for precise, high-speed industry analysis rather than relying solely on Text2Cypher.
- Multi-Hop Depth: Capable of 4+ hop reasoning (Director -> Movie -> Genre -> Decade) to provide deep historical context for every answer.
Screenshot of my agent in action:
Link to my agent (Github): GitHub - Kumar3421/CineGraph-AI · GitHub