Live from Confab: Metadata is a love note to the future

Metadata is a Love Note to the Future, by Rachel Lovinger

Metadata is in code, it’s in security, it is context, and it enables connections.

It’s hard to convey that in a concise and powerful way.

Rachel’s fiance (ooooooh!) had a slide “metadata is a love note to the future” which has now made the rounds via retweets, Tumblr, other peoples’ presentations… and now it’s come full circle.

Rachel had to take ~10 years of magazine content (about 50k articles) and put them into a CMS. 3 years later, doing a redesign, they found almost 7000 different keywords (though they found ~12% were typos, or redundant keywords).

But they wanted the metadata to do more.

The Goal: immersive experiences

• If you like James Bond, you should be able to easily discover everything they had.
• Allow people to get from James Bond to Daniel Craig and so on
• Cross-over between media types (book reviews, movie reviews, DVD reviews)

What they did

• They created a hierarchical system of terms.
• They standardized the metadata relationships, (i.e. if it’s tagged with X, it automatically appears as related content on all other pages tagged with X)
• The relationships were defined and managed separately from the article content
• They built a “category tool”

The results

• Better search results
  • For example, searching “Green mile” gave you the option of “The Green Mile (book)” or “The Green Mile (movie)”
  • User can see an annotated list of every article tagged with one option or the other, when they choose one.
• People found tons of older content!
• Other people wanted to use the categorization database
• Showing every related item is not scalable – it shows way too much information.

Challenges of metadata

• It is a lot of work
• Is it accurate?
• Is it credible?

For an archive, Jason’s team at archive.org does detailed metadata:

• Media type is automatically detected, and tagged
• Additional metadata is manually entered
• The attributes are determined by the community, which is a dedicated community – however, a lot of user-generated metadata is inconsistent or poor.

Metadata gives future people (sometimes future us!) information to find the things they need.