Data for companies

1. What the dataset is

ValorScore maintains a catalog of thousands of movies, series, games, books, and albums, each scored across 14 content and values axes — the same data that powers the product you see on the site.

Just as IMDb licenses audience ratings and JustWatch licenses streaming availability, we license our values-evaluation dataset to companies that want to integrate this layer of information: recommendation apps, streaming platforms, parental-control tools, content aggregators, and market research.

2. What is included

Per title: the 14 consolidated axes (0 to 10), the three aggregates — Woke Score, Virtues, and ValorScore —, the IMDb rating where available, and basic metadata (type, title, release date). Per-axis evidence (the scene, theme, or excerpt backing a score) can be included in plans that cover it.

The dataset is versioned (every export carries a header with the schema version, generation date, and row count) and distributed as JSONL or CSV, depending on your pipeline's needs.

3. Schema (per title)

Each dataset row follows this format:

• id — Internal title UUID
• type — Vertical (movie/series/game/book/music)
• title — Title
• releaseDate — Release date (ISO 8601) or null
• imdbRating — IMDb rating (0–10) or null
• valorScore — ValorScore — composite 0–100 "worth watching" score
• wokeScore — Woke Score — 0 to −100 (ideological magnitude)
• qualityScore — Virtues — 0 to 100
• axes — The 14 consolidated axes (0–10 each)
• evidence — (optional) per-axis evidence: refId/scene/quote

4. Public sample

Before discussing licensing, you can download a limited public sample with evidence stripped: /downloads/dataset-sample.jsonl. It shows the exact format of the full dataset, featuring the titles with the highest ValorScore.

5. Licensing model

We license the dataset by subscription (access to periodic exports) or as a one-time snapshot, priced by title volume, update frequency, and intended use (internal vs. redistributed to end users).

Permitted use: integration into your own product, internal analysis, research. Attribution required ("Rating data by ValorScore — valorscore.com") whenever the data is shown to an end user. Redistributing the raw dataset to third parties, reselling it as a standalone data product, and using it to train competing content-evaluation models require a separate agreement — get in touch.

As with the rest of the platform: scores are EDITORIAL OPINION WITH EVIDENCE, produced by AI and community — never a factual verdict, an official classification, or an aggregation of third-party sources. Licensees adopt this same framing when displaying the data to their users; see the full methodology at /en/metodologia.

6. Get in touch

For a licensing proposal, a larger sample under NDA, or schema questions: dados@valorscore.com. Tell us your use case, estimated volume, and desired update frequency.