The following is Part 2 of an article by Abby Clobridge that was first published in Taxonomy Times, a publication of SLA, in November of 2015. Read Meaningful Metadata: Part 1.
Descriptions Matter
Longer, free text description fields help can certainly help with findability and information retrieval. For repositories using Dublin Core as the metadata scheme, take advantage of the “dc.abstract” and “dc.description” fields. When articles or manuscripts are deposited into a repository, this field is particularly important — but be careful to follow copyright restrictions if the article does not include a CC-BY or CC-BY-NC license that allows for re-use as in many cases, publishers own copyright.
Longer, free text fields, whether a formal abstract or simply a longer description of the item, can also help to improve an object’s performance in search results. Richer descriptions using multiple words and phrases in natural language can provide context — something that computers need in order to help determine the relevance of an item when returning search results.
Incorporating Metadata into Multiple Places
Another useful practice for improving information retrieval is adding metadata to multiple places. All repositories include metadata fields — at a minimum, some sort of title, author, and date fields, even if those are automatically populated by the system. SharePoint includes “Dublin Core columns,” and all Open Access repository systems include fields based on Dublin Core, MARC, or another commonly-used metadata schema.
But in addition to the formal metadata records within a repository system, each digital object has the potential to include additional metadata as well. Items saved in Microsoft Office (Word, Excel, PowerPoint) or PDFs all can include additional metadata. Furthermore, for public-facing repositories, heading tags matter — incorporating good metadata into headings (i.e., <h1>Metadata Matters!</h1>) helps with information retrieval.
In all of these cases, it is important to incorporate key phrases and words in an organic, meaningful way. The frequency words and phrases, and their proximity to other words, all matters.
User Tags Plus Taxonomic Terms
Within records, if possible use a mix of taxonomy terms through pick lists or drop-down menus, free-text tag fields, and longer free text description fields. In SharePoint, it is possible to use a drop-down/pick-list that includes a space for “other” and an associated free-text field users can use to define additional categories. Mining that data can help inform what terms might be considered to be added to an existing taxonomy or spotting trends, emerging issues, or hot topics — which might be added to a taxonomy or used in other ways to aggregate and push content to users, content that might be invisible without a curator’s intervention.
Show Your Metadata
By forcing metadata to the forefront through breadcrumbs or tag clouds can help with discoverability. By making taxonomy terms or categories clickable, users can serendipitously find other related items — seeing additional categories might prompt users to click to find other materials they might not otherwise have discovered.
Think Globally
If appropriate, use multiple languages for selected portions of metadata. Titles, taxonomy terms/subjects, and descriptions are areas where multiple languages could potentially have the most impact. Know your particular audience to determine whether multiple languages might be useful or beneficial.
Metadata: Not Just a “One and Done” Effort
Most importantly, don’t think of metadata work as a one-time effort for a particular item or record. Continuous improvements and enhancements to metadata help keep records fresh. Apply the lessons learned by search queries to enhance records. What are the top keywords that are drawing people to your site or repository? Are these keywords included in all appropriate records? Words, phrases, and concepts — along with search algorithms — change over time. Make sure your metadata records reflect these changes.
