# Migrate from OpenAlex or Semantic Scholar

**Outcome:** your existing metadata pipeline keeps working, with grounded answers added on top.

## Field mapping

| Their field | Ours | Notes |
|---|---|---|
| `id` / `paperId` | `id` (`pap_…`) | different namespace; map via DOI |
| `doi` | `doi` | the reliable join key |
| `title` | `title` | |
| `abstract` / `abstractInvertedIndex` | `abstract` | plain text, not inverted |
| `tldr` | `tldr` | |
| `authorships[]` | `authors[]` | flatter shape |
| `host_venue` / `venue` | `venue` | typed `journal` \| `conference` |
| `open_access` | `open_access` | |
| `cited_by_count` | `citation_count` | |
| — | `is_retracted` | no direct equivalent; use it |

Verify each row against the Paper object before relying on it — field names here follow the v1 design and the OpenAPI spec is authoritative.

## What you gain

Grounded answers with page-level citations, extraction over your own PDFs, and 2,000+ citation
styles — none of which have an equivalent in a metadata-only API.

## What changes operationally

- Auth becomes a key with scopes rather than an anonymous or polite-pool request
- Calls are metered in credits — budget before you migrate a bulk job
- Licensing terms differ; re-read content-licensing before rehosting anything

## Related

papers · content-licensing · pricing-and-credits