Tagging, or social bookmarking, emerged in the internet this past year through popular sites like del.icio.us and Flickr. It allows people to put metadata (labels) on content, primarily internet links in the case of del.icio.us and photos at Flickr. Tagging is the "offspring" of keywords, but with some new twists. Anyone can tag anything any way they want; there is no agreed-upon or imposed taxonomy. In addition, multiple tags to the same object allow bookmarks to belong to more than one category, bypassing a limitation of the traditional hierarchically organized category systems.

There are other differences between tagging and conventional classification. Readers—not just writers and librarians—get to tag. The new tagging systems are web-based, so they can become accessible to all, and for these two reasons tagging becomes social. This social quality also allows taxonomies to be built from the ground up by users, rather than be determined by designated experts.

FREE TO BE
The total freedom of tagging has some classification purists upset or dismissing it as chaos that will collapse under its own expanding weight. Some argue that tagging should be done only with a controlled vocabulary. Others support giving taggers complete freedom to develop their own keywords. If you leave the choice of tags open, it is likely to initially be a bit chaotic for each topic, but then standards and common terms will start to emerge as the majority of people choose the more popular terms. This creates a "bottom-up" fluid taxonomy, sometimes called a folksonomy, rather than an imposed one. In practice, this fluidity can allow for the constantly changing way we look at things to emerge.

Social bookmarking is one of the more recent examples of emerging new internet tools, sometimes called Web 2.0, which can have very useful applications within an intranet. These tools share the concepts of transparency and access to search, while each is optimized for specific functions. Social bookmarking is optimized for documenting and sharing lists of links to useful information. In contrast, blogs are optimized for creating and sharing content, allowing for commentary while keeping the original content intact. Wikis, on the other hand, are optimized for the collaborative creation of content.

Taking social bookmarking behind the firewall opens up new uses, including the following:

• Providing topical resource lists that can be personalized and shared—creating personal "knowledge management" systems through lists of winning sales proposals, best practice deliverables, etc., around specific topics or work efforts;
• Extending individual profiles to let others know what content an individual considers important;
• Facilitating discovery of employees with similar interests or facing similar issues;
• Offering support to online workgroup activity;
• Measuring popularity of intranet documents through numbers of tags; and
• Supplementing enterprise search engines through the emergence of new keywords that are meaningful to employees.

SOCIAL WORK
IBM's Dogear provides an excellent example of an enterprise social bookmarking tool in action. The developers from IBM Research, David Millen, Jonathan Feinberg, and Bernard Kerr, optimized Dogear for intranet use, resulting in tagging capability combined with authentication of the tagger. You can tag intranet as well as internet material, and your corporate-directory presence can have your tags (as can your blogs). You can see others within your firm with the same bookmarks as well as the bookmarks of individuals. and you can subscribe to them through RSS. IBM added collaborative filtering to infer stuff that people might want from their tagging activity and let them know about it. IBM has also integrated its enterprise social bookmarking system with its corporate search engines, giving the user access to the benefits of both.

How does tagging fit in the evolution of enterprise taxonomy development? According to Millen, in the past, metadata came from several sources: First it was created by professionals, often from the library sciences, and then through automation. Through social bookmarking, "everyman" is asked to engage in tagging (and through it, keyword development), but unlike predecessors, their efforts are aggregated and made transparent by technology. Millen feels that one of the reasons this has been successful is that the tagging structure is visible to the user in a useful way, rather than hidden, as it was in the past.

Adopting social tagging within the enterprise creates new demands, as well as new opportunities. For example, the potential chaos of tagging is less desirable. According to Millen, Dogear will likely evolve to better support consistency and reduce the potential chaos in tagging through such enhancements as making synonyms and alternative spellings visible to the user. Currently there is an auto-complete feature that encourages consistency. Another option being considered is to recommend recent and popular tags. They are also looking at how social bookmarking taxonomies map to corporate taxonomies to discover ways to improve corporate classification systems.

In their pilot of Dogear, the developers used social networking analysis to look for patterns of shared interest. They found several clusters of individuals: those who are seekers of information and those who are information providers. This social nature of tagging provides another data source to help devise ways to better share information within the enterprise. Dogear has been demonstrated to IBM clients with a very positive response, and a number of them have asked to be part of any early product deployment.

PLAY WELL WITH OTHERS
As social bookmarking moves into the enterprise environment, additional features will be added to enhance its usefulness to organizations. Dogear is not alone in adapting to user's needs. Another example of expanded features can be found in the work of the H20 Project at Harvard's Berkman Center for Internet and Society. It has developed a social bookmarking tool, the H20 Playlist. This tool was inspired by Wikipedia, iTunes, and del.icio.us, and it offers a shared list of readings and other content on a topic of intellectual interest. This could be a course syllabus but is certainly not limited to this use. As Playlist developer (and "Senior Geek in Residence") Luis Villa notes, "Playlists provide a tool optimized to enable the sharing of ordered lists of information. It was designed with enough features to be useful and not too many to make it difficult to use by busy professionals."

To achieve their goals, Playlists have more structure than del.icio.us, including headings and subheadings. The Playlist can have a title separate from its tags, which makes it a unique entity that can be shared through its URL. At the same time you can add multiple tags to a single Playlist to retain this tagging benefit. You can also reorder your Playlist to reflect changing priorities instead of being limited to reverse chronological order in del.icio.us. Like iTunes, a person can set up a library of all their links, and then create multiple Playlists that draw on these links, adding to the flexibility. You can also see all the Playlists that contain an item, as well as all the comments that others have made about the item. When looking at a Playlist you can also see links to other Playlists:

• by the same author
• influenced by this one
• derived from this one
• with the same items
• with the same tags

All these features add to the usefulness for enterprise taggers and promote further knowledge sharing and discovery of people with common interests. Villa sees Playlists operating midway between blogs, where one person generally owns the content creation, and wikis, were multiple people continuously update the same content. With social bookmarking tools like Playlists, the creator serves as editor: filtering, ordering, and annotating content created by many others. Like blogs, the creators are identified and their work remains intact. Like wikis, the content can come from many contributors. It can also be reused in multiple ways with the ability to trace its origins back to the source.

Playlists have the potential to work within an enterprise as well. A knowledge manager can create an ongoing library of links to critical documents with annotations on their importance. Then official enterprise Playlists can be developed that represent the best thinking on specific issues that are important to the company. Employees can download Playlists to fit their work needs. Then they can make these Playlists personal as they engage in work activities by modifying them with new annotations on the usefulness of existing links as well as new links. Anyone in the firm, including the original knowledge manager, can access these new derived Playlists to enhance their own or add back into the official enterprise Playlist on the topic. The ability to stay current is facilitated by the Playlist feature that allows you to see what Playlists are being derived from existing ones, as well as what Playlists are influenced by others.

With tagging behind the firewall, the best thinking in the firm can become more transparent and be constantly updated to the benefit of all.