schema.org is a shared vocabulary for describing things on the web, such as people, places, creative works, and actions. Chronicle borrows its model and names: where schema.org has a term for something, Chronicle uses that term's name and points its own term to schema.org's.
What it's for
Web publishers add schema.org to their pages so that search engines and other programs can tell what a page is about, such as an article, a recipe, a product, or an event. It covers hundreds of kinds of things. Its rules are loose: a property lists the types it expects, but other values are allowed, and no property is required.
Why Chronicle has its own vocabulary
Search engines read schema.org on web pages and make the best of whatever they find. Chronicle's records are written by many independent plugins, kept in a personal archive, and read by programs years later. That needs three things that schema.org leaves open on purpose:
- One way to write each fact. Two plugins that see the same thing must write the same record, or one thing becomes two. schema.org accepts several forms for most values, such as plain text where it expects a person.
- A closed set of terms. "What did I watch in 2019?" has a complete answer only if every plugin records watching the same way. Chronicle has about a hundred classes, each added when a source needs it. schema.org has over 900, and anyone can add more.
- Meanings that stay put. A record kept for years must still mean what it meant when it was written. Chronicle versions its vocabulary and publishes each release, which it can do only for terms it owns.
Restricting schema.org's terms to meet these needs would still make a second vocabulary, one that uses schema.org's names but means something narrower. A program reading schemaorg:width would expect schema.org's meaning, not Chronicle's number of pixels. And schema.org's strength, that any program reading a web page understands it, matters little for records that stay in an archive. So Chronicle defines its own terms and points each to the closest schema.org term, for translating data as it enters or leaves an archive.
Shared model
Chronicle describes what happened the way schema.org does. An Action has an agent who performed it and an object it was carried out on, and it can have a result, an instrument, a target, and a location. A work is a CreativeWork, with an author and a datePublished. A media file is a MediaObject, with subclasses such as VideoObject.
Most Chronicle classes sit under the same parents as the schema.org classes they are named after. Each class and property points to the closest schema.org term, as a hint for mapping data. The hints don't make a Chronicle VideoObject a schemaorg:VideoObject: Chronicle's values follow its own rules, and some terms differ in scope or structure, as the notes below say.
What Chronicle adds
- Where each record came from. Every record has a source and a sourceId, and a key built from the source's own identifiers, so the same thing gets the same identity each time it is read.
- What happened in one person's history. Chronicle has actions that schema.org does not, such as CompleteAction, ExecuteAction, MessageAction, PublishAction, and VisitAction. It also has periods of life, such as a Tenure or a DeviceSession.
- Stricter values. Each property lists the types its values can have and whether it takes one value or a list, and records that break either rule are rejected. A DateTime can be a date known only to the year, or an uncertain one.
- A ledger. Money that moved is a FinancialTransaction, with a FinancialPosting for each FinancialAccount it changed, so an archive can keep accounts the way a bookkeeper does. schema.org describes money as commerce: a schemaorg:MoneyTransfer is the act of moving it, a schemaorg:Invoice is a bill, and a schemaorg:BankAccount is a product a bank offers. None of them records one movement of money or the accounts it changed.
What Chronicle leaves out
Most of schema.org. Chronicle adds a term only when a source needs it, so it has no terms for selling, such as offers and prices, or for the many kinds of web pages and organizations that schema.org describes.
Translating a record
To read a Chronicle record as schema.org, replace each term with the schema.org term it maps to. A term that is the same as a schema.org term, or a kind of one, keeps its meaning. A close match needs the note beside it. For example, a width is a number of pixels, and the object of a TravelAction is where the travel began, which schema.org gives as schemaorg:fromLocation.
Some schema.org names mean something else in Chronicle, such as Project and QuoteAction. The tables below mark these as not the same. Chronicle's own terms, such as sourceId and VisitAction, have no schema.org counterpart.