The benefits to information users of having content tagged with a taxonomy are significant: increased accuracy, comprehensiveness of search results, etc. However, often the greatest challenge to taxonomy adoption is the ability to tag all of the content with the taxonomy terms as intended. Issues include allocating resources for tagging, implementing a new content management workflow, establishing criteria and quality control for tagging, and tagging a large volume of legacy untagged content.
While taxonomy development has one-time project expenses (such as the hours of consultant or contractor), the ongoing tagging with a taxonomy requires an annual budget on top of some startup expenses, whether tagging is manual or automated. Manual tagging requires budgeting for the working hours while auto-tagging typically requires an annual software license. Automated tagging also requires some human involvement for quality checks and refinements of tagging parameters.
Automated methods are more cost-effective for large volumes of content tagging and can tag more quickly. Automated (AI) methods can tag text or images, but the same tool/technology does not do both, so for mixed content, manual tagging may be a more practical and affordable option. Automated methods are also better for the content of a consistent type (e.g. all resumes, all news, all technical support articles), whereas a diversity of content (e.g. everything on the intranet or the public website), can be tagged more accurately if done manually.
Automated tagging is not free from manual labor. If tagging is done by machine learning, then the machine needs to learn from examples, and sample tagged content may need to be prepared and submitted to the system as such examples. If tagging is done by rules, then rules need to be written for most of the taxonomy concepts. Prebuilt starter taxonomies may be pre-trained or have tagging rules included, though, but they likely will need refinement. Any auto-tagging needs to be tuned and refined as the content and the taxonomy evolve.
Establishing criteria and quality control for tagging begins with setting tagging policies and guidelines. This includes setting the policy regarding what detail to tag, how many terms of each type may be tagged to a single piece of content, whether a certain taxonomy term type is required or not for tagging, and whether the tagging of certain terms should trigger the additional tagging of another term (such as a broader term). These policies can be set as parameters for auto-tagging. For manual tagging, some of the tagging policies can be system enforced, but other policies cannot be.
Even if there is an established workflow for tagging newly added content, there is the challenge of tagging all the legacy content already in the system. A taxonomy is rarely implemented before any content is collected and available for searching. While taxonomy creation is a project, taxonomy management and maintenance are an ongoing program, and it’s the same with tagging.
Click here to read the original article published by Hedden Information Management.
Please give your feedback on this article or share a similar story for publishing by clicking here.
