Choose the keywords to look for
Choose which words the keyword model looks for in your photos: add, move and remove words, sort them into categories and set how sure it must be for each.
Open the keyword editor with Edit keywords… under Machine learning. It opens in a large window of its own, called Keywords.
The board
Each category is a column, and its words are chips beneath it. The columns scroll sideways, and each scrolls on its own. Find a word at the top narrows the board to matching words.
Words
- Type in Add a word… at the foot of a column to add one.
- Drag a word by its text to move it to another column.
- Hover a word and click its ✕, or press
Delon a focused chip, to remove it. - Each chip has a small knob: its own cut-off, the score the model must give before the word counts for a photo. Unset, it follows the library's default. Drag it up or down to give the word a cut-off of its own, and click it to put it back to the default.
Categories
- New category adds a column at the end of the board.
- Double-click a column's name to rename it, and drag it by its name to move it.
- Removing a category that still has words asks where they go: Move them to another category, then Move and remove, or Remove the words too.
Checking the list
Check for near-duplicates looks for words that mean nearly the same thing. Each close pair is listed, closest first, and the chips involved carry an amber dot.
Restore the predefined list puts the list back to the words and categories it shipped with, after asking.
After an edit, the library's photos are scored again in the background.