Methodology

What the dataset is

The dataset separates observed CBS title rows, clustered public game records, enriched canonical metadata, and audit files.

What was excluded

Utilities, editors/SDKs, guide media, and disc-noise rows are removed from the canonical public game catalog but remain available through audit files.

Match states

• Matched: a canonical match survived the audit rules.
• Ambiguous: multiple plausible targets remain or the match was too weak.
• Unmatched: no trustworthy canonical target is published.

Data quality score

`data_quality_score` is a heuristic summary combining content class, merge confidence, canonical match state, link presence, and match demotions.

Audit files

• schema_map.md
• improvement_log.md
• excluded_non_game_titles.csv
• match_demotions.csv