In 1990, I was introduced to a way of annotating files and documents with tags when our office got a copy of Lotus Magellan to share on one of the PCs. We put all of our documents and client correspondence into it and added tags and context to help us find things later on. The advent of the Web brought similar solutions for tagging content, but it was a bit of a mess. Delicious sorted out much of the mess by adding identity into the mix of tags on objects, which provides better insight into the context of the tag and its meaning. In 2004, I coined the term folksonomy to separate the new way of tagging that included the person who tagged the item and sharing, which made a big difference by allowing broader and clearer use of tags.
But how does tagging fit one's personal knowledge management. Our human fuzzy brains can benefit from tagging, but they also make tagging a wee bit flakey. Personal tagging efforts (shared or not) have some good uses such as:
- aggregating similar items by topic, subject and/or for use;
- refinding things easily when search is limited or across systems and services;
- adding missing metadata or terms; and
- adding emergent vocabulary.
There are four things to consider when tagging: 1) What type of object is being tagged? 2) Are you tagging across systems and services? 3) How good is your search? 4) How easy to use is the interface at relative scale?
Starting with the type of object, it is often helpful to have tagging on media objects that have no annotation or limited transcription (if any at all). Text documents can be tagged or not with subject matter and or categorical type tagging. This depends on how good the search is for the system they are in.
Audio, video challenges
Audio, images and video are trickier because they don't have natural text hooks in them. Images with good annotations and metadata may benefit still from tagging. Audio and video are more problematic because the meaning and content are time-related, and it is much more difficult just to scan a service to understand what lies within the media. Audio and video optimally should permit tagging of the timeline as well as tagging the media as a whole object, but that is not the norm. Services like Viddler and Videojuicer, open source services to build a video service, include tagging and other annotations within video timelines. For audio, SoundCloud (soundcloud.com) offers annotatable timeline tracking entries by person.
All of the previous services are Web-based, so aggregating one's own tags across the services is a bit more difficult. One means, if the services allow for it, is to get a feed for each of your tags and pull them into a feed reader or other RSS feed aggregation service with search. Being able to search one repository for "cytoplasm" makes things a lot easier, even if that search has pointers back out to other services to fully get to the content or media.
The search factor
Search is another dependency that will influence how much, as well as how, you tag things. Where search is weak or poor, tags are going to be your best buddy. Where search is good can often lead to tagging things differently for aggregation. The use of tags for missing metadata or context that isn't clear becomes the norm. Where this has a down side is the cross service and platform tag aggregation. Good search often reduces the tags applied over time.
Lastly, the interface where objects are tagged has a lot to do with how and what types of tags are added. This really became relevant when comparing people's Flickr photo repository to their iPhoto desktop photo repository. Flickr often had a subset of the people's iPhotos repository, but Flickr was more heavily tagged. When people looked in iPhoto, they could scroll through time ordered thumbnails to quickly find what they were seeking in short order, but Flickr has photos on pages that are not easily scrolled, so to quickly find and aggregate images used tags more heavily.