project Utilities / File Location Listing avatar

utilities/file_location_listing#53: Should images inherit tags?

Issue Information

Issue Type: issue
Status: closed
Reported By: btasker
Assigned To: btasker

Milestone: v0.2.7
Created: 31-May-24 16:27



Description

Images don't currently get associated with tags, and so won't show up if a tag specific search is used.

This can lead to a confusing experience:

I search for corsa:

Screenshot_20240531_172446

Then I click the corsa tag under the first result to make sure we're being more specific

Screenshot_20240531_172537

The image suggestions are gone.

When we index images, they sometimes inherit from the page that's anchored them (or at least, the first one we find)

        if image.has_attr('alt') and len(image['alt']) > 0:
            linkinfo['title'] = image['alt']
        else:
            # There's no alt text, so use the pages title (if available)
            linkinfo['title'] = title

Should we perhaps also inherit the parent page's keywords?



Toggle State Changes

Activity


31-May-24 16:27

assigned to @btasker


31-May-24 16:33

It's relatively easy to envisage scenarios where this would be helpful.

However, it relies on pages only containing images that relate to a single topic, and that's not always true.

For example, in my most recent blog post, I've got

  • Grafana graphs showing solar usage and yield
  • Grafana cells showing amortisation times
  • Grafana cells showing savings/usage during octopus plunge pricing events
  • Cells showing octopus pricing buckets
  • An image describing powerups

The post itself carries the keywords solar, electrical, octopus, homeassistant, house stuff, blog, analysis.

Under this proposal, each of those images would be associated with each of those tags. So, searching for homeassistant would start returning graphs about solar as well as powerups.

I think that there's a good chance that it would lead to a significant drop in the quality of results being returned.


31-May-24 16:41

We could perhaps add a per-site config option to control this behaviour - that way domains that do closely tie tags + images (photography domains for example) could have tag inheritance toggled on.

The only thing with that, is the inconsistency in the way that we currently collect and associate images.

If you've got a blog-roll type category view enabled, we may crawl that before the image's "main" page and so wouldn't associate tags (or a title) - first page to be seen embedding currently wins.

I think it's worse to return incomplete results than no result - at least the latter gives a sign that something's not right.

Closing as Won't Fix - if we ever get as far as redesigning how we link embedded pages, this could be reexplored.