Tagging Classification

The following is an excerpt of “Tag, You’re It! Best Practices for Tagging on the Web” from the January/February 2010 issue of Law Practice Magazine:

“…Given that it’s [tagging] a personal classification system, it naturally follows that there aren’t many fixed rules when it comes to tagging. Consequently, you’ll be given every opportunity to mess it up when selecting your keywords and phrases. Fortunately, that’s perfectly acceptable because it’s your use of language that needs to be represented. Still, to make your tags more user-friendly, you need to find ways to minimize the potential chaos. Here are some tips.

?First and foremost, be consistent. Your primary objective is to be as consistent as possible in applying tag terms. Before choosing terms, try to resolve some of the fundamental language issues. For example, will you use singular or plural words? Will you restrict your tags to nouns, or are verbs okay? Will you use capitalization or stick to lowercase letters? Once you develop your initial set of terms and find a preferred style, stick to it!

?Don’t replicate terms. Remember, tagging is designed to fill gaps where traditional categories leave off. When tagging a blog post, for example, try to find terms or phrases that are relevant but not mentioned in the body of the post. Avoid repeating the same terms, or applying different stems (e.g., “-ed” or “-ing”) to the same word.

?Keep them short. Some tagging systems only allow single keywords, while others let you string together short phrases. When the latter is permitted, target two- to three-word phrases as a maximum. Any longer and you greatly diminish your chances of reapplying the same tag elsewhere.

?Alter your numbers based on media vs. content. The number of tags you apply will depend on what you’re classifying. With images such as digital photos, for example, there are no written words to search, so reasonably, photo or image collections deserve to have a higher number of tag terms applied. As many as 10 tags may be appropriate to convey both the content and the concepts represented by the image. By comparison, when tagging content items or links, it’s best to limit the number of tags applied. This is particularly true for blog posts, so you don’t replicate terms already embedded within the item. Blog posts are often well represented with two to four tags in total.

?Limit use of abbreviations. Make sure abbreviations or acronyms are recognizable by your intended audience (even if the audience is only you), and consider potential conflicts in how they’ll be applied. Law librarians, for example, might avoid the tag “ ALA,” since in addition to being an acronym for the American Library Association, it also represents the Association of Legal Administrators.

?Tag trends, products, personal names and organizations. Think in terms of the “Five Ws”—who, what, when, where and why—or play 20 questions with yourself if you must, but identify the most recognizable elements when selecting terms.

?Consider the search engines. Niche topics have smaller audiences, and more differentiated terminology. But hand in hand with that, any time tags are applied to a public Web collection, term aggregation pages will show up in the search results and become new entry points into the collection. Selecting tags based on your audience’s familiarity with the terms can be an important consideration in making your collection accessible.

?Revisit your choices. Last but not least, a periodic review of your style never hurts. A good general strategy is to let tagging happen naturally for a couple of months and then look back at the terms you’ve applied. If you’ve got 13 items tagged as “snowboard accidents” and two as “snowboarding accidents,” your preferred style is clear—and a small intervention may be in order to make your tags consistent going forward.

In our opinio, sometimes is preferable to retrieve information using concepts (highprecision) instead of key-words (low-precision).


Leave a Reply

Your email address will not be published. Required fields are marked *