MARC mostly just took data from cards. We have strings of text with inconsistent punctuation. Extra data, such as inferred dates and subject headings.
Designed for description rather than discovery. Catalogs tell you if the library has a specific edition of a work, but doesn’t pull together different editions very well.
His site: MARC usage in WorldCat
The Legacy Problem: inconsistent usage, such as formats in 245 $h.
The Title Problem: All the different editions and forms of “Hamlet” in one big list with no distinctions.
The Name Problem: John Rock, the abolitionist or the scientist.
Solving the problems: name all the things (“entity”). Define relationships with other entities.
Triples: A person called Albert Einstein wrote a work called Relativity about Physics.
Persons: wikidata, VIAF
Works: WorldCat Works
Subjects: LC Subject Headings
These are not *records*. They are assertions. Each can stand alone. A record is a set of assertions.
Google tells you how to use linked data on your page, but linked data is not search engine optimization.
Site: Linked Jazz.
WorldCat Linked Data: WorldCat Works, Fiction Finder
(Solving the Title Problem)
Could have a page on an author that pulls in bio from wikidata, book info from WorldCat, etc.
WorldCat Entity Cards
U. Of Wisconsin catalog: queries VIAF, then goes to other sources, such as Getty, DBPedia, etc.
OCLC is linking a work to its translations.
Authority data: VIAF and other identifiers in Wikipedia. (Solves the Name Problem)
Legacy data needs work. Legacy systems need work. Bibframe isn’t ready. A steep learning curve.