Recommended Keyword Practices

Tagging, aka keywording, an image collection using a digital asset management (DAM) system or image management tool is a big topic! How it is done is hugely contextual; highly dependent on the given industry and the company within that industry. That said, here are a number of best practices to consider which should be of immediate help and provide food for thought on how to do it best for your scenario. While some are DBGallery specific, many are generic practices.

Avoid keywording individual images where possible.

Starting tagging the selecting the largest number of images possible at one time. If tagging a folder or new set of images, select all those and tag them with common attributes. Then within those, select a small group with common attributes and tag them. Tag individual images only when they’re unique.

In General

Consider if only admins can add new keywords. This has the benefit of a cleaner list of keywords and prevents inappropriate or incorrect data from being entered. (In DBGallery, use the Keywords Dictionary tool and choose “Fixed keyword dictionary”.) Note that within DBGallery all data for an image is searchable. This includes folder and file names, exif and iptc metadata, dates info such as the year, and collection names. Hence perhaps some keywords aren’t needed, such as data already in the folder name, or the camera and year. Consider marking images as having data tagged or not. This can be via a special tag itself, or even better, creating a DBGallery workflow: New Image, Partial Data Entered, Data Complete. (These are defined in the Workflow tool within DBGallery.) Consider having new or images not yet tagged placed in a non-published folder that only admins or data entry personal have access to, then move them to a published area for general access once tagged. Given a large collection, consider enlisting help. This can be work-term students or volunteers. Availability of volunteers is heavily organization specific, such as parents volunteering to tag school photos or your clubs membership tagging images.

Use AI Auto-tagging

While it won’t fully replace manual tagging anytime soon, it can be a huge help reducing the amount of manual tagging.

Pro: can be helpful in that they are effort free and can provide keywords, descriptions, etc. you wouldn’t normally key but are useful in a search.
Con: can return results you wouldn’t expect. There can be incorrect keywords. At lease try AI auto-tagging for a while. You will almost surely find that it greatly reduces manual tagging. Misc tags it adds that you likely wouldn’t bother with can be useful. And in the end if the cons outweigh the pros, all it’s data can generally be easily removed.

Also see our blog posts Heaps of Automated Metadata in DBGallery and DBGallery’s Metadata Capabilities for deeper thoughts on this topic.

Specific keywords

Consider tagging the dominant color. Can be useful when matching colors with a web page or printed page.
Tag location info where relevant to an image. If images have GPS stamps, DBGallery will auto added address info.
Add a mood when that is a prominent aspect of the image.
Use plural spelling only, unless the spelling is different than just adding an “s.” For example, just using “cats” is sufficient, but if you use “babies” also include “baby.”
If a collection is accessed internationally, list alternative spellings for international audiences (centre / center, gray / grey, color / colour.)
Use the subject field. This can be an important tag where users may be looking for landscapes, portraits, medical, food, and other subject. Use multiple subject items where appropriate.
Consider your audience when tagging. What do those searching this collection look for in images.
Other good tips on what data to enter for an image can be found here, where some of the above suggestions were taken from: http://www.controlledvocabulary.com/metalogging/ck_guidelines.html

Importing Keywords

Other than manual and AI tagging, there are additional ways to import data:

All images added will have existing metadata read and added to the system. This include Exif and IPTC/XMP metadata.
Add images and data by uploading a csv file and associated images in a single zip file. This allow for any number of data fields to be imported. While especially useful when initially populating a collection, it can be used at anytime to added sets of images and their data. (See DBGallery’s Data Import Guide)

Searching

Be sure to provide search tips to users. No matter the amount of data entry effort, suggesting how to find images can be key to a making a collection more valuable. Let them know about wildcard, exclusion and inclusion searches (*, +, - respectively). For additional search tips within DBGallery, see our Smarter Search blog entry or Help on the main menu. Provide a channel for those searching images to have a way to provide feedback on tagging. This can be huge in ensuring images can be found by all users even if sometimes combersome given too much or shallow feedback. Missing or misspelled keywords can be identified, whole categories of keywords or other data can be suggested. Offer search and image collection navigation training from time to time. Making sessions available for later viewing.

Custom fields

Consider creating custom data fields tailored to your organization’s specific needs, such as descriptive, technical, and administrative metadata. This can be exceptionally useful and make it extra easy for those searching that have specific data they’re looking for such as Project Number, Cliet ID, SKU, etc. It being its own field makes it stand out from a data entry and data view perspective. For how to create them, see Custom Data Fields.

Conclusion

While most companies would want to have a perfect tagging plan at the started, the truth is that unless there is a good deal of company-specific image management experience available on the team for that plan it’s likely to be something which evolves over time. Presuming that isn’t the scenerio where you work, the above recommendations are good place to start.