Tagging: People-Powered Metadata for the Social Web

I recently read Tagging: People-Powered Metadata for the Social Web (2008), by Gene Smith. Smith dives into tagging as a method for adding metadata to resources, which in turn increases the organization and findability of the resources.

Traditional help authoring tools categorize resources through folders (a carryover from Windows folders), whereas web platforms typically use tags. Tags are actually a quick and easy way to attach metadata to any information object.

For example, you might tag a photo with a geolocation. This would allow the photo to appear in the correct location on a map. Or you might tag a help topic with information about the audience and other relevant facets, such as role, location, goal, task/concept/reference, and so on. The key point is the tags are metadata.

Twitter wouldn’t be nearly as powerful without tags. Adding #techcomm to a tweet makes it findable for the technical communication community, just as tagging tweets with #contentstrategy makes it findable for the content strategy folks.

What’s cool about tags is that you can easily tag the same information with multiple tags. Does the information fall in #techcomm, #contentstrategy, and #findability disciplines? You don’t have to choose one folder to assign the content to. With tags, resources can live in multiple places at once.

Tags can be generated in at least two different ways. If users assign tags to topics, the result is a more free-form, loose set of terms that some call a folksonomy. Del.icio.us is the flagship example of a folksonomy. If only designated authors can assign tags, the result is more of a centralized taxonomy.

Tag clouds, which are visual displays of your top tags, can give you a general idea of topic trends. For example, if you look at the bottom of my sidebar, I frequently tag posts with the terms blogging, creativity, podcasting, screencasts, STC, technical writing, Web 2.0, and WordPress.

The tag cloud highlights only the most common terms. To see 500 tags in a cloud, see this page. When you look at this massive tag cloud, you might immediately realize one of the limitations of tags: lack of hierarchy. Most tag systems lack any kind of hierarchical arrangements (that is, parent and child tags), so as tag systems grow, they become unwieldy. The flat structure of links becomes hard to navigate.

Managing tags in digital photo collections can highlight the limitations of tags. Scott Dart, program manager of the Microsoft Photo Gallery, says,

“Anyone who has tagged their photos for any length of time will tell you that a flat list eventually becomes unwieldy. This is one of the reasons why we have hierarchical folder structures– because a flat list of folders would be too long to manage” (quoted in Tagging, 200).

In other words, the ease of adding tags to resources leads to an abundance of tags that gets to be hard to manage, unless you have some type of hierarchy imposed on your tagging systems.

Some tag clouds do allow you to drill down and explore a list of subordinate tags. Other platforms allow you to leverage tags in combinations of each other. Both strategies can make tags more useful and powerful.

Beyond tag hierarchies, Smith talks about some concepts I hadn’t heard before:

• Pivot Browsing: ”Moving through an information space by choosing a new reference point — a pivot — for exploring the system” (105). The ability to look at information based on users, resources, and tags provides various pivot points in the data. Pivot points seem similar to facets but more radical in the way they change the reference point.
• Pace Layering:  ”Aspects of society change at different rates” (91). For example, tags suit user needs as they’re moving quickly to make sense of new information; taxonomies and ontologies are more appropriate later, when the dust settles and users have more time.
• Synonym Rings: “A synonym ring gives two or more words an equivalent meaning” (69). These rings establish synonym equivalents for tags, so that “Web20″ and “Web_2.0,” for example, would be equated with one another.

Some tagging systems present users with the most popular tags, often segmented by time. But knowing that a tag is popular is just one type of information. Popular by whomis another type of information that tagging can surface.  Smith writes, ”One of the benefits of tags to object-oriented sociality is to bring people together through their tags” (187).  In other words, you can also connect with other users who have tagging patterns similar to you.

Tagging interfaces are characterized by speed and simplicity. As a result, tagging can be messy, with numerous tags having similar spellings, formatting, and synonyms. This is why it’s necessary to regularly clean up tags with tag management tools, which allow you to merge, delete, or change tags in bulk.

Tags have been implemented in a variety of ways with different platforms. For some examples, check out Del.icio.us, Mefeedia, Buzzillions, LibraryThing, Milenix MyInfo, Flickr, and Youtube.

Tagging is emerging as one of the most common ways to organize resources on the web. Smith notes that Delicious, a social bookmarking site that first appeared in 2003, was one of the first instances of tagging to emerge online. However, he also notes that classification and metadata strategies have been ongoing for centuries, with the Dewey Decimal system as one of the prime examples.

My Thoughts

In the larger discussion about findability, tagging brings us right back to the metadata discussion. Tagging is metadata that people apply to resources so they can find the resources later.

Tags present a radical shift in the way we attempt to organize information. Many help authoring tools default to hierarchical folder structures, but tags allow for polyhierarchy and give many more “pivot points” and facets for browsing the information.

The key point about tags is that you can add any number of them to a single resource. You can then manipulate the resources based on the tags you want to leverage. This is something you can’t easily do with a traditional hierarchical organization of information, or in systems where resources are placed in one folder at a time.

For example, with help information, you might add tags related to any of the following metadata properties:

• Author
• Date published
• Release version
• Date last revised
• Popularity
• Task
• Concept
• Reference
• Format
• Difficulty
• Location
• Event

You could then manipulate the resources in different ways based on the metadata that was important to you. You could also manipulate the resources based on combinations of tags — for example, the most popular tasks that are specific to a particular location. Combining tags provides a powerful way to sort and manipulate data, as it allows for a lot of different arrangements and possibilities.

When we start thinking of tags as metadata, they becomes a much more useful tool for help systems.

Although I’ve explored different strategies for findability, it seems that faceted classification through the attachment of metadata (such as tags) to resources remains the most compelling strategy. It can suit a diversity of audiences, purposes, and needs.

Taking it one step further, I think we need to allow users to tag help content. One failure of tech comm is that it hasn’t kept step with the innovation of the web. Many help authoring tools discard the interactivity of the web and the wisdom of the crowd. The revolution that needs to occur to pull help into the current era is to leverage the wisdom of the crowd in an intelligent way to increase the findability of help. Allowing users to tag content, and then leveraging their tags, seems like a good way to start.

Blog Sponsors

• Webworks ePublisher
• Scriptorium
• Help Generator help authoring software
• Southern Polytechnic: Information Design and Communication
• Simplified English
• MindTouch
• Madcap Software
• Dr.Explain
• Adobe Technical Communication Suite
• Congree

Faceted Classification

In a faceted classification system, you tag your content with a specific set of attributes. You then provide different arrangements and sorting of the content based on those attributes.

For example, if you have an online paint store, the set of attributes for your faceted classification system might include brand, color, texture, shine, thickness, and cost. You can then allow users to navigate the paint products by each of these attributes.

Sarah Maddox notes that in Matthew Ellison’s Turning Search into Find presentation at the AODC 10 conference, he gave an example of faceted classification with the shoe store sportsshoes.com.

Faceted classification system

Many times faceted classification systems allow you to combine and narrow down the attributes to get closer and closer to your information. Sarah notes that

Matthew introduced the concept of the “scent of information”: If people can see that they’re getting nearer to the information that they’re want, they’re quite happy to keep combining facets to narrow down their search.

Faceted classification and faceted search systems are actually quite common on the web. When you search for something on Google, the left pane provides a list of facets that you can use to limit the search results to a specific category of information.

Google's faceted search

In the above example, I searched for “undercover agents.” Using the facets, I can limit the results to show images, videos, news, discussions, shopping sites, maps, books, or other formats that contain the words “undercover agents.”  This is the way Google has categorized the different types of information it indexes.

Amazon also provides a faceted search. When you search for a product, such as Beauty and the Beast, you see a list of facets in the sidebar to narrow the results.

Faceted browsing on Amazon

Grooveshark also provides a faceted search based on the term you search for.

Grooveshark's faceted search

Navigation Features

The most popular sites on the web may not all be set up with faceted classification and faceted search, but they do offer other types of navigational features. These other navigational features could informally be called facets, but according to Peter Morville, author of Search Patterns and Ambient Findability, these organization systems are more like features rather than facets. But they still organize and arrange the content into structures that aren’t based on hierarchical topic containers. Here are a few examples.

Hulu allows users to choose among the following:

• Gallery content
• Channels
• Most Popular
• Recently Added
• Collections
• Trailers
• Spotlights
• TV
• Movies
• Search

Wikipedia provides users with these navigational features:

• Featured content
• Current events
• Random article
• Today’s Featured Article
• In the News
• On this day
• Today’s featured picture
• Categories
• Index

The New York Times provides these features for navigating the content:

• Today’s Paper
• Video
• Most Popular
• Times Topics (like an index)
• Categories
• Most E-mailed
• Most Blogged
• Most Viewed
• Most Searched
• What We’re Reading
• Reader’s Recommendations
• Most Recent

Coming back to Amazon, when you view a product on Amazon, you get a ton of interesting navigational features for finding more content:

• Customers Viewing This Page May Be Interested in These Sponsored Links
• What Do Customers Ultimately Buy After Viewing This Item?
• Customers Who Bought This Item Also Bought
• Tags Customers Associate with This Product
• Customer Reviews
• Customer Discussions
• Look for Similar Items by Category
• Your Recent History

You can see the wide variety of faceted classification, faceted search, and other navigation features on the web. Why don’t we borrow more of these faceted navigation models to provide alternative organization systems for our help content? Why is it that, despite the abundance of these alternative navigation systems, most technical writers still stick with a traditional topic-based, hierarchical folders? In contrast to many sites on the web, the organizational structure of most online help files hasn’t changed in 20 years.

Blog Sponsors

• Webworks ePublisher
• Scriptorium
• Help Generator help authoring software
• Southern Polytechnic: Information Design and Communication
• Simplified English
• MindTouch
• Madcap Software
• Dr.Explain
• Adobe Technical Communication Suite
• Congree

Ann Rockley

A few years ago, I wanted to better understand content management, so I picked up Managing Enterprise Content, by Ann Rockley, and read it through. It opened my eyes to a lot of new concepts.

Ann is one of our field’s leading experts in content management. She’s now expanding in to something she calls “intelligent content.”

Intelligent content is a concept that builds on other concepts you may already be familiar with. I think we’re going to hear a lot more about intelligent content. In fact, she and Scott Abel are preparing an entire conference exploring intelligent content. I caught up with Ann over email and asked her to expand on the concept. Below is our interview.

What is intelligent content?

We define it this way: Intelligent content is structurally rich and semantically aware, and is therefore automatically discoverable, reusable, reconfigurable, and adaptable.

Let me explain more what that means.

Structurally rich

“Structurally rich” means the content is structured content, and more importantly it is semantically structured content. For example, we can look at the structure and know what type of content it contains (steps contain chronological action-oriented information). DITA-based content is an example of structurally rich content, as is DocBook, XBRL, and RSS, though the content could have a custom structure as well).

If we have a structure in our content, we can manipulate it. For example, we can automatically determine how to publish it to multiple channels (print, web, help, mobile), or we can filter out some content (e.g., tables may not work as well in the mobile environment).

Also, if it is structurally rich we can perform searches and narrow our search to the particular type of information we are interested in (e.g., look for all occurrences of the word metadata in conceptual information).

Semantically aware

The word semantic refers to “meaning.” Semantically aware content is content that has been tagged with metadata to identify the kind of content within it. For example, you might tag your content with industry, role or audience, and product. If the content is tagged with semantic metadata, it is possible to automatically build customized information sets based on audience or industry, for example.

As more organizations start to create personalized content (content which is dynamically assembled upon user request that specifically matches a users need or user profile), this type of metadata becomes extremely important.

In addition, as content is pushed to wikis, integrated through “mashups” or “pipes,” it becomes even more important to ensure our content is semantically tagged. Without semantic metadata, it’s difficult to automatically, let alone manually, find the content we need.

Discoverable

If the content has semantic tags and is structurally rich, it’s a whole lot easier to find exactly what we are looking for.

Reusable

Reusability refers to content we frequently use. If content is structured for reuse, and I know what type of content it is, I can either easily retrieve it for manual reuse or automatically retrieve it for systematic reuse (automatic reuse).

Reconfigurable

Knowing the structure of the content, we can output it to multiple channels, reconfiguring it to best meet the needs of the channel, or we can automatically mix and match content to provide us with the information the customers needs. We can even transform content (reconfigure it) from one structure to another, but only if we know what the structure is in the first place.

Adaptable

We frequently create our content for a particular need or audience, but content can be adapted (used in a different way), often without our knowledge, to meet a new need.

Is “intelligent content” a new term for our industry? If so, who coined it?

As far as technical communication goes, “intelligent content” is a new term. In some ways, it’s a new term in the broader content industry as well.

I coined the term, just like I did for much of the terminology used today for reuse because there wasn’t a term to describe something that existed, or there were too many terms, and talking about something or trying to explain something was difficult.

Technical communicators are very focused on producing high quality content that meets the customers’ needs, often in a very short period time and often with tools that won’t stretch to meet their needs. Many have begun to move to DITA and some are adopting content management, but when you have the conversation with management about why they should move to DITA and adopt content management, it is very difficult to get across the concepts and the return on investment. DITA is a standard, content management is a tool, but how does it help the organization to do what they need to do better?

I’ve heard some managers respond to a well-presented business case with, “So, why should I care, how does this really help us?” Let’s turn it around, let’s talk about the goal and what it gives the organization and the customer. Intelligent content allows us to do the following:

• Automatically publish to multiple channels because the content is structured. Content structure can be recognized and automatically transformed to any format we like. This is not possible with traditional content.
• Customize our content for customers because we can identify what content is appropriate for what customer.
• Reduce the costs of translation because content is modular and designed for reuse, etc.

Why is it intelligent? Because it is structured, etc.

How do we support intelligent content? With DITA, component content management, etc.

Now let’s take the concept of Intelligent Content outside of Technical Communication. There are huge amounts of information in organizations being provided on websites. For a long time, metadata has been the way that most companies optimized content for retrieval, but now XML is beginning to make inroads into broader organizational content, and that brings all the benefits I’ve already discussed.

If you try and talk about XML, though, you’ll lose most managers because it is perceived as being too technical. However, you can turn it around and say we can create intelligent content that enables us to:

• More easily find it
• Deliver it
• Customize it
• Personalize it
• Automatically deliver it to multiple channels
• Simultaneously release content in multiple languages

And

• Reduce costs
• Speed up delivery time
• Optimize resources
• Do more with the same resources
• Increase customer satisfaction

We’re creating intelligent content that can be automatically discoverable, reusable, reconfigurable, and adaptable, etc. — we’re not just creating XML-based content.

What’s the role of the content creator in creating intelligent content?

The role of the content creator is crucial in intelligent content. It is not enough to just put our content in topics and push it out, although that’s a good start. We need to think about all the ways in which we can make our content adaptable. This means doing content analysis, customer needs analysis, and identifying an appropriate information architecture that sits above our content. Content creators are getting a good handle on DITA and structure, but very few use or understand metadata.

What kind of tools do you need to create intelligent content?

We can use the existing DITA tool sets and Component Content Management systems, but if we are interested in helping the organization beyond Tech Pubs, we should consider using XML content servers, and dynamic delivery engines.

What skills are needed for a writer to create smart content?

I’m always looking at where writers are today and where they can go to increase their skills and marketability, so this list reflects a growth curve. To start, writers should look at DITA, but in looking to the future they should gain an understanding of the following:

• Content modeling
• Metadata
• Reuse strategies
• Multichannel delivery strategies
• Mashups, pipes
• XPath and XQuery (not necessarily so they can code these, but so they know what they can do and specify requirements for them)
• Structuring content for marketing campaigns (the intricacies of campaigns are incredible and they are completely dependent upon intelligent content)

What organizations are employing tactics to make content intelligent?

We are seeing the beginnings of intelligent content with organizations that have moved to DITA, but organizations that are global or that have a broad product lines are really developing intelligent content. They are tagging content for region, product, audience, and more as well as automatically producing content that meets the needs of their customers. We have one client that creates 500 different pieces of content to reflect different products from the same content source, all automatically and all based on metadata tagging and DITA!

Next we are seeing intelligent content in organizations where their product is content (e.g., newspapers, magazines, publishers).

We are also seeing intelligent content in pharma, medical devices, intelligence, and financial industry organizations.

Ann Rockley is part of The Rockley Group, a content management company focused on intelligent content.

For more information about Ann Rockley, see The Rockley Group. You can also follow The Rockley blog.