DEV Community

DropThe
DropThe

Posted on

We Built a Knowledge Graph With 2M+ Connections. Here's What It Reveals About Movies.

We maintain a database of 1.83 million entities — movies, TV series, games, companies, people, cryptocurrencies, universities — connected by 2.18 million knowledge graph links.

One of the things we built on top of this is the Feelgood Score: an algorithmic scoring system that rates 191,000 movies on comfort, not quality.

The stack:

  • PostgreSQL database with 1.83M entities
  • JSONB data columns per entity (ratings, genres, sentiment, streaming availability)
  • Custom scoring algorithm weighing audience sentiment, rewatchability, genre patterns, and critical reception
  • Knowledge graph links connecting actors to films to studios to streaming services

What the data shows:

When you score 191,000 movies on comfort instead of quality, the rankings flip completely. The Godfather (rating: 8.69) scores 29 on comfort. Zootopia (rating: 7.76) scores 80.

The correlation between audience rating and Feelgood Score is nearly zero. Quality and comfort are independent variables.

Some stats from our movie database:

  • 209,684 total movies tracked
  • 113 languages represented
  • $820.7 billion in total tracked box office revenue
  • Only 1,126 films confirmed profitable (5.5% of those with financial data)
  • 685,896 streaming availability records across 678 services

We publish all the stats openly: Movie Statistics 2026

Full writeup on the Feelgood Score analysis: We Scored 191,000 Movies

The scoring methodology is transparent: How We Compute the Good Score

If you're building anything with large entity databases or knowledge graphs, happy to talk architecture.

Top comments (0)