Tag: tagging

Algorithms and tools are groping around in the dark, with only the limited tool of tagging to help them figure it out. Let’s say I want to share a blog post on Facebook. I drop in my link, and the page handily populates with information on the post:

All this draws from the page’s metadata and feeds into Facebook’s Open Graph algorithm that determines what the best headline, intro description and image are. If you’re expecting others to share your content, setting up the metadata to feed them the right information will be key – so your copy is the right length and the right image gets pulled in and linked.

Creating Your Tags

When you’re thinking about creating tags, consider which types are most appropriate:

• Descriptive – Terms like #ocean or #beach that say something about what’s in the image, or meta tags that describe the content on the page.
• Image type (for images only)– Qualities of the picture itself – a close-up, a landscape, a soft-focus image.
• Contextual – Relates to the conversation that you’d like to be in – becoming a part of that discussion.
• Conversational – When the tag becomes the conversation. This most commonly happens on Twitter, where hashtags such as the joking apology of #sorrynotsorry are more message than meta.

The Future of Tagging

Search engines like Google have moved away from keyword tagging and towards automatically analyzing the text and structure of a webpage in order to draw conclusions. Similarly, as image processing gets more advanced, algorithms are able to parse out some of the details of what’s in an image.

As an example, Shutterstock recently launched an auto-tagging tool for its mobile image uploading. The tool looks through the existing metadata/tags in the current library of images, maps it against the content of the image, and provides a range of suggestions. For a nature photo, these might include “flower,” “nature,” “beautiful,” “red,”closeup” and more. As these tools become more prominent, we should expect:

• Clustering – We’re likely to see more of what already exists. If people already know to search for #destinationwedding in order to find content related to weddings, we’ll see more and more uses of that tag by people who want to show up in that context, and the tools will continue to recommend it.
• Tags substituting for descriptions – Descriptions are challenging to write, since they need to encompass all that a piece of content contains. Tags are easy, since they can be single facets of that content, because they’re automatically recommended, and because they automatically feed search engines. Expect to see the continued growth of numerous tags over lengthy descriptions of content.

Where do you see the future of content tagging?

Quick, describe Raphael’s St. Michael and the Dragon in 12 adjectives or less. Would they include #knight #religion, #greatart and #chiaroscuro? Would we be missing some of the essence of the image in boiling it down to these tags – these simple, searchable snippets?

The digital world is a tagged world, a world coded and snipped into little boxes. Content must be deconstructed into its essential elements and coded in this way so that the algorithms that curate content for us (Google, Facebook, etc.) can put them into the appropriate boxes. It’s most obvious on channels like Instagram, where an image might have 10 or more hashtags coding it:

It’s also apparent in many other contexts, such as meta tags on webpages to improve their search engine optimization (though Google and other search engines have moved to de-emphasize them in their ongoing algorithm updates in favor of content- and link-based analysis).

But it’s not intelligently curated, and it doesn’t speak to quality. I can tag any image or page anything, without that necessarily implying that it’s actually related or that it’s going to be relevant. Even if my tags are accurate, what something is isn’t always what it’s about; content, whether visual or text-based, doesn’t make sense without its context the unspoken relationships it has with other concepts matters deeply in understanding it. English, as with most languages, is very context-oriented. If I say “spring” to you, I could mean:

• Spring (the season)
• to spring (the verb)
• “Spring!” (the verb as a command)
• a spring (a water source or an elastic object)

Without additional terms, it’s near impossible to know which one is referred to.

Content without curated context

We have unprecedented flexibility in the ways we sort, filter and understand the world online. Yet this poses a new challenge once we come out the other side and work to understand the content. In the physical world, content tends to be placed within a certain context by its curators. To get to the Raphael paintings, you walk through galleries of his predecessors’ art, and to find a book on robotics in the library, the Dewey Decimal system places it with other robotics books.

Online search engines, social channels and other electronic middlemen let us tag things in dozens of different ways and then search based on them, such as subject, color, author, data format, production date, organizations or people mentioned, or more. The results then show up based on that search – putting each item in a context it wasn’t necessarily intended for. Just as with “spring,” if I search for #ocean, I could find the romantic image above, or an image of storm-tossed ships on the verge of destruction, or an article on marine biology – with nothing in common other than this single aspect. Each result must stand alone and be interpreted alone.

We’ve asked algorithms to be our curators, helping us find what we need in whatever way we’re thinking about it. This is an immense opportunity to draw new connections and find new content. Yet the challenge is that to make this possible, we must squash down content into a few tags for search, then try to re-expand it on the other side into its full richness. The more we can emphasize that richness while still making it possible to find, the more likely our content is to resonate and earn results.

Read more on implementing content tagging and the implications of auto-tagging in Tagging Content for Users and Algorithms.