Attachment metadata for advanced searching

The idea here would be to have a way to have a note that acts as metadata for another file type.
(Of course this could be done manually)

For example images could have a written description of the contents of the image. PDFs could have an index of the text in the doc.

The note could show all back links to the file.

Could be interesting with some automation/AI - images could be automatically tagged with objects in them for example.

6 Likes

This would be a great PKM advancement. It would allow users to use the same app for both image- and text- based mindmapping/ mood boarding without the hack of manually embedding images into .md files. Adding pdf metadata would allow the app to address the text format that is most difficult to convert to well-formatted markdown using current tools.

I would love this. I am planning on using Obsidian as my universal document store, for everything I scan as well as related notes and accounts etc, and the ability to add searchable metadata for imported blobs would be amazing.

Personally all I want is a flexible tag and k/v store. So the ability to add just random words to an object like “finances”, but also have “created by” : “morgan”.

Cheers!

I do this manually or with help of AutoHotkey scripts.

I create text file with the same full filename and add suffix “.txt”.
e.g. “article.pdf” + “article.pdf.txt” or “videoplayback.mp4” + “videoplayback.mp4.txt
In Windows Explorer they display next to each other when sorted by name. When searching for original filename, the metadata file shows in results as well.
Automatic checks for existence of paired file can be performed just by adding/trimming “.txt” suffix to the path. The same to open it.
Similar method is officially used/supported to associate folder with html file by adding “_files” suffix to filename like “a.html” file + “a_files” folder.

Now with Obsidian, I might switch from “.txt” to “.md” suffix but I do not want to. reasons:


Edit: also requested in Link attachment to page it describes

1 Like