CORPUS · 4,726+ INDEXED · 141 COUNTRIES · AND GROWING

Our data.

A 7-year structured film evaluation corpus

PACCS is trained on 4,726+ real film submissions across 141 countries — drawn from active festival programming, not scraped trade-press scores. Every record carries metadata, narrative signals, audience indicators, and outcome labels.

4,726+
Films submitted
141
Countries
7y
Of evaluations
12
Outcome labels per film
UK / EU
Region-locked hosting

Curated from 7 years of independent festival programming. All evaluations anonymized.

SHAPE OF THE CORPUS

What's in the data.

LAYER · 01

Submission metadata

Title, director, country, runtime, genre, fest-circuit history, platform availability — structured and normalised.

LAYER · 02

Narrative & visual signals

Pacing classification, dialogue density, visual language profile, score signature — extracted from poster, trailer, and synopsis.

LAYER · 03

Programming outcomes

Festival fit, acceptance, audience response, distributor pickup, territorial release patterns — labelled across 12 outcome dimensions.

LAYER · 04

Expert evaluation history

7 years of structured evaluations from international festival programmers — the curation logic PACCS learns to recognise.

PROVENANCE

Where the data comes from.

Submissions flow through international festival channels we operate or partner with — not screen-scraping. Every row is consented for analysis under our published data terms. Hosting region-locked to UK/EU. Audit retention is permanent.

Consented submissions UK/EU hosted Permanent audit retention Published data terms Source citations on every score