--- title: Working with Artifacts description: Edit artifact metadata — titles, dates, creators, people, places, subjects, and rights — the museum-standard way. section: guides order: 4 updated: 2026-07-16 verified: 2026-07-16 related: [reference/dates, guides/collections-and-permissions, guides/ai-analysis, guides/subjects-concepts-and-authorities, guides/identifiers-and-numbering] features: [artifact-metadata, places-and-maps, creator-person-linkage] --- # Working with Artifacts You'll learn what each artifact field is for and how to describe your material in a way that powers browsing, search, and attribution. An **artifact** is the intellectual object — the photograph, letter, or recording itself. The digital files behind it (the scan, an access copy, a transcript) are **assets** attached to the artifact. You describe the artifact; the files take care of themselves. ## Core fields - **Title** — what visitors see everywhere. Descriptive beats clever: "Loggers at the Bloedel-Donovan mill, circa 1925" outperforms "Mill photo #3". - **Identifier** — your accession number. Free-form; shown on the detail page and searchable. - **Type** — image, document, audio, video, or object. Usually set automatically from the first file. - **Description** — rich text (Markdown) shown on the artifact page. This is the main field AI analysis drafts for you. - **Date** — supports the full messiness of historical dates, from "March 15, 1923" to "circa 1920s" to "after the flood". See [Historical Dates](/docs/en-US/reference/dates) for what you can type and how it's interpreted. ## People: creators vs. who's in the picture Preservated separates two questions that catalogs often blur: - **Creators** answer *who made this* — the photographer, artist, or donor, each with a role label. Creators appear in credit lines and citations. - **People** answer *who appears in this* — individuals depicted, mentioned, or identified in the artifact, linked to Person records. Person records accumulate every artifact a person appears in and power face-recognition suggestions. When face detection finds a person in a newly imported photo, it proposes a match to an existing Person; confirming it links the artifact for you. ### Creators link to person pages When a creator is linked to a person record, their name on the artifact page becomes a link to that person's page — their biography, other works, and any photos they appear in. The [PastPerfect import](/docs/en-US/guides/importing-photos#import-a-pastperfect-export-folder) populates this link automatically, matching every creator name to a person record by exact name match; a creator name from other sources shows as plain text until it's linked. A person's page lists every artifact linked this way under **Works by**, separate from the **Appears in** list of artifacts they're depicted in — a photographer who also appears in a group portrait shows up in both. You can also link, unlink, or create a person from a creator by hand. In the artifact editor, each row in the Creators list has a person control next to the name: - **Unlinked** creators show a small link icon — use it to search for and link an existing person, or to create a new person record using the creator's name exactly as typed. - **Linked** creators show the linked person's name as a badge — open it to unlink (the typed name stays; only the connection is removed) or to view the person's page. Linking or creating never changes the name printed on the artifact — that stays whatever you typed. If the linked person's own name is later edited and no longer matches, an **Update display name from person** button appears on that row; it only shows up when the two names differ, and only copies the person's name onto the creator when you click it. **How names are shown.** A person's name reads in natural order — "John Smith" — in credit lines, face labels, and on the public person page, but the People list and browse views file it by family name — "Smith, John". You can enter a name either way: Preservated derives both forms and keeps the original exactly as you typed it. Names it can't confidently split — single names, some non-Latin scripts, or an organization or family entered in the People field — fall back to showing what you typed and are flagged for review in the People admin list. ## Subjects **Subjects** are topical tags ("logging", "schools", "shipbuilding"). They drive the browse facets, so consistent terms pay off. Every descriptive term carries a **role** (object name, subject, genre/form, material, and more) and can link to a controlled vocabulary. See [Subjects, Concepts, and Authorities](/docs/en-US/guides/subjects-concepts-and-authorities) for how roles, concept links, and local terms work, and [Authorities and Controlled Vocabularies](/docs/en-US/reference/authorities) for what each vocabulary is for. ## Places and maps **Places** are the locations tied to an artifact. Type a name and pick the matching record from GeoNames, the Getty Thesaurus of Geographic Names, or Wikidata; Preservated stores that authority's coordinates and a clean display name alongside the label, so the place can appear on maps and in standards-based exports. A plain typed label works fine when no authority fits — a place is never required to be linked. Any kind of named place can be linked, not just cities: bridges, parks, lakes, channels, mountains, cemeteries, schools, and other named features all match. When results with the same name differ in kind, each non-city result shows what it is ("Bridge — Washington, United States") so you can tell a bridge from the strait it crosses. ### U.S. street addresses When what you type looks like a complete U.S. street address — a house number, street, and either a ZIP code or a city and state — a **U.S. addresses** group appears above the authority results, powered by the U.S. Census Bureau's public geocoder. Picking one stores the standardized address as the place label along with its coordinates, so the artifact can appear on maps. Two things to know about address places: - The point is **interpolated along the street's address range**, not measured at the building — close enough for a neighborhood map pin, but not survey-grade. - An address has no authority record, so it's stored as a local place *with* coordinates. It behaves like any other place in search, maps, and exports. PO boxes and non-U.S. addresses can't be located this way; enter them as plain labels. ### Display names Picking an authority record gives you two names: a concise **label** for chips and lists, and a fuller **hierarchy** for hover and detail. Both are formatted for your institution's home region — a US place trims the country ("Barron, Washington"), while a place abroad keeps it ("Vancouver, British Columbia, Canada"). Re-pick the place if you want the names recomputed. ### How places get coordinates - **You pick an authority record.** Its point — and, for a region, its extent — comes with the chosen result, along with the preferred display name. - **AI reads a location off the image.** When [AI analysis](/docs/en-US/guides/ai-analysis) reports something like "Barron, Washington", Preservated resolves the most-specific name against the authorities and checks it against the surrounding context (state, county, country). A confident, unambiguous match is applied for you and tagged as an AI suggestion you can revert; an uncertain match is proposed for review; and a genuinely ambiguous one is kept as a plain label rather than binding the wrong record. A location is never discarded. Catalog-style qualified headings are understood too — "Deception Pass Bridge (Wash.)" resolves to the bridge in Washington, and a qualifier like "(bridge site)" steers the match toward the right kind of feature when a town shares the name. Headings that name no place at all ("Statewide", "Unknown") stay plain labels. - **A heading is a U.S. street address.** Imported or AI-reported labels that read as a complete street address are looked up against the Census geocoder instead; a single confident match is always proposed for your review — never applied automatically — and an ambiguous one stays a plain label. - **Photos with embedded GPS** get capture coordinates automatically at import. On the artifact page, a place with coordinates is shown on a map when the displayed file carries no GPS of its own (a file's own capture/subject coordinates are more precise and take precedence). The map zooms to match the place's precision — a street address sits close in, a whole state pulls back. Places that carry coordinates are also published in the artifact's IIIF manifest as map geometry (`navPlace`), so IIIF-aware viewers can plot where the content is. The **depicted** location — where the photo's subject is — is preferred over the camera's position. ## Rights The rights block records the copyright story: a human-readable statement, a license URL (Creative Commons URLs are recognized), the rights holder, a credit line, and any institutional restrictions. From these fields Preservated computes a **copyright status** (public domain, likely public domain, unknown, likely under copyright, under copyright) per jurisdiction, with its reasoning shown to visitors. That status gates what visitors can do — download originals, remix, order prints — so accurate rights metadata directly controls feature availability. :::tip For CC-licensed material, paste the license URL rather than describing the license in prose — the URL is what the permission logic reads. ::: ## Provenance: who changed what Every edit is recorded in the audit log with structured before/after diffs. AI-generated suggestions are stored separately from your curated values and never overwrite a field without an explicit accept — you can always distinguish "the AI guessed this" from "a curator confirmed this". ## Where the rest lives - Organizing artifacts and controlling who sees them, including staff-only attachments: [Collections and Visibility](/docs/en-US/guides/collections-and-permissions) - Letting AI draft metadata: [AI Analysis](/docs/en-US/guides/ai-analysis) - Transcripts and OCR text for documents and recordings: [Transcription and OCR](/docs/en-US/guides/transcription-and-ocr)