At the start of this month, I released the first beta of a new app that reverses Spotlight search to build a vocabulary of keywords found in your images and other documents. I’m delighted today to release the second beta of that app, Spotcord, with many significant improvements.

The background to this is how to improve our use of metadata to find the documents we want. One general approach is to add keywords, then search for one or more of those. If I wanted to find an image showing sheep with their lambs on a fine day, I might try searching in Spotlight for images with keywords containing lamb and blue sky. This of course depends on whether any images have been assigned those keywords. If none use the keyword lamb, then widening the search to sheep might return too many hits that don’t include lambs. If you’re going to use keywords, then knowing the vocabulary is essential.

As far as I’m aware, neither Spotlight nor any software using it can compile such a vocabulary, at least not until Grant suggested that I might write Spotcord. I’m also very grateful to John who got me to look at how macOS Ventura and Sonoma perform their image analysis but don’t populate keywords for images.

Currently, almost all keyword metadata are set by the user. Some apps now offer the ability to analyse images for you and generate a list of appropriate keywords that you can then save in their IPTC/EXIF metadata. This is one of many useful features in GraphicConverter, for example.

Keywords are also available in PDF, Word docx, Pages RTFD and many other formats, where they have to be added manually in their document information editor.

When you’ve got a folder that you want to build a vocabulary of keywords for, open a new document in Spotcord, and set it to scan that folder.

The app always scans Keywords (kMDItemKeywords), and you can add to those Subject (kMDItemSubject), and any metadata type supported by Spotlight. Its scan is deep, including every file and folder inside that selected. In this case, it included 116,800 files in all, and found 3,542 keywords across them. Its vocabulary is sorted alphabetically (sorting by frequency is coming in a future release), gives the frequency of each keyword, and can be saved to a text file.

This is version 0.8, which I’m hoping is getting close to its first full release. In addition to the new features I’ve just described, it now has a nine-page Help book that even lists suggestions for custom Keys.

There are two significant limitations imposed by macOS. I’m afraid that it can’t build vocabularies for images in Photos libraries, as those aren’t accessible as regular metadata, but stored in databases instead. As my excursion into image classification in Ventura and Sonoma showed, it also can’t build vocabularies for objects automatically recognised by macOS, as those don’t work that way. But for IPTC/EXIF and all those other keyword metadata, it works as advertised, and picks up data embedded in the documents or attached using extended attributes.

Spotcord version 0.8, its second beta, for Big Sur and later, is now available from here: spotcord08
from Downloads above, and from its Product Page. I’ll be adding it to my auto-update system when it reaches full release.

Happy keywording!