Structured data is common and essential in many situations. Despite prolonged attempts to structure text semantically, most that is indexed and searched has mark-up removed, and uses the lowest common denominator, plain text.
search
One JSON text file consistently found a search term that another contained, but could never find. How do some documents get fully indexed, but others don’t, and can’t be searched?
Spotlight’s search window provides ‘AsYouTypeTopHit’ ranked search, including a measure of recency, 3 opaque scores, related search, and hits across multiple search domains. Here’s how it does that.
From the outset, Spotlight has included websites and other items in ranked search, and local files listed exhaustively using filters. As we work with ever more data, ranking is becoming more important in our search methods.
Spotlight’s search window is incremental and ranks search results across many domains. How this works in practice, and why it may not be ideal when searching for files.
Two apps using Core Spotlight, compared with respect to indexing of new entries, and searching. Quick tests to tell whether an app uses Core Spotlight, and what that brings.
When listing files by name, should 20test.text come before or after 200test.text? Which order is the more ‘natural’? How this quickly becomes complex.
The phrase repeated throughout WWDC was ‘semantic search’. How does fit in with Spotlight, and how could it benefit those who aren’t so enthused by advanced AI?
Spotlight can’t index the contents of document versions to make them searchable. How you can change that and save having to browse those versions when you need to recover old content.
You realise that a few hours ago you trashed an important file by accident. How can you search your Time Machine backups without looking through them one at a time?
The Eclectic Light Company 