# imdb-tsv

Trimmed IMDb TSV snapshots, packaged for direct load into the graph schema
defined in `schema.cypher` (Kùzu-style `CREATE NODE TABLE` / `CREATE REL TABLE`).

## Source

Downloaded on 2026-07-07 from the official IMDb dataset mirror:
<https://datasets.imdbws.com/>

The three source files used were:

- `name.basics.tsv.gz`   → `name.basics.trimmed.csv.zst`
- `title.basics.tsv.gz`  → `title.basics.trimmed.csv.zst`
- `title.principals.tsv.gz` → `title.principals.trimmed.csv.zst`

## Processing

The official IMDb TSVs are tab-separated and contain extra columns beyond
what `schema.cypher` declares. Each source file was **lightly processed**
(header kept, columns projected to the ones the schema references,
field names renamed to match the Cypher property names) and recompressed
with `zstd` (`.zst`). No row filtering, no value rewriting — the row
counts and order match the upstream dump.

| File                          | Rows (incl. header) | Source columns kept                          |
| ----------------------------- | -------------------: | -------------------------------------------- |
| `name.basics.trimmed.csv.zst` |           15,468,443 | `nconst`, `primaryName`, `birthYear`, `deathYear` |
| `title.basics.trimmed.csv.zst`|           12,623,654 | `tconst`, `primaryTitle`, `startYear`, `runtimeMinutes` |
| `title.principals.trimmed.csv.zst` | 100,343,239  | `nconst`, `tconst`                            |

Column renames applied during trimming:

- `name.basics`: `nconst → nconst`, `primaryName → primaryName`, `birthYear → birthYear`, `deathYear → deathYear` (no rename; just dropped `primaryProfession`, `knownForTitles`)
- `title.basics`: `tconst → movieId`, `primaryTitle → title`, `startYear → year`, `runtimeMinutes → durationMins` (dropped `titleType`, `originalTitle`, `isAdult`, `endYear`, `genres`)
- `title.principals`: kept `nconst`, `tconst` (dropped `ordering`, `category`, `job`, `characters`)

The result is the minimal subset needed to materialize the `Actor`, `Movie`
and `ActedIn` graph from `schema.cypher` via the COPY statements in
`copy.cypher`.

## Loading

Decompress the `.zst` files (e.g. with `zstd -d`) and then run
`copy.cypher` against a database whose schema has been created from
`schema.cypher`:

```bash
zstd -d *.zst
# create schema from schema.cypher, then:
# run copy.cypher
```

## Provenience / licensing

Data © IMDb and its licensors. Use is subject to IMDb's non-commercial /
personal-use terms for the datasets dump:
<https://www.imdb.com/interfaces/>
This repository is a redistribution of a snapshot of that data; please
respect the upstream license when using these files.