Basic Keywroding Concepts

Keywords: keywording?

At some point, most photographers, even the most casual ones, will have to deal with indexing and locating specific images from the many they’ve taken over the years. Simply sorting them into folders by date or location can quickly become problematic especially when you have 10,000 images, let alone 10s of thousands. What do you do when you want to find a picture of a Snowy Owl shot on a specific beach but have thousands pictures taken at the same place over 10 years? The answer is searchable metadata and specifically keywords. Even if you don’t intend to pursue photography seriously, there’s a good chance in 10 years you’ll have quite a few snapshots you’ll want to be able to  find things quickly.

The best time to address this is before it becomes a problem. That is the sooner you develop a solid keywording system the better.

What makes good keywords?

For our purposes, a keyword is a noun, verb, or adjective that generally describes the content, behavior, style or some other key factor of an image. A good keyword is also reusable. That is it can be used on many different images not just one or ones from a single shoot.

For example, “people”, “family”, “building”, “car”, or “Green Heron” are good keywords, “Cousin Tom Smith” generally isn’t, because you can’t apply it to anything other than Cousin Tom Smith. As a matter of convention, I won’t  use peoples’ names as keywords, even if I photograph them regularly.

When I’m creating keywords I’m always asking myself can I use this keyword on another image that doesn’t include this specific subject.

Those things that make poor keywords often can be expressed better somewhere else in the images metadata. Usually this is done though EXIF and IPTC standard fields intended for location, job, scene and title, among others. These fields should be used as well whenever possible for that information in place of keywords. In addition some fields are automatically used by online photo galleys and photo-sharing sites to automatically generate titles, captions and provide geographical information place images on maps or similar.

I only place location related information into the IPTC location block not in keywords. This includes scene, location, city, state, and country information for where the image was taken. In addition, I deal with people’s names by identifying the people in the image’s caption. This has the advantage of allowing me to be more expressive (“the people in this image from left to right are…”) as well.

Beyond that it’s important to remember to be consistent and that keywords are there for your benefit and for the benefit of people searching though your images (if you share them online). There are no right or wrong keywords. However, there are good keywords and not so good keywords

Beyond Flat Keywords

The most basic approach to keywording is to simply develop a flat list of keywords. In fact, most basic image editing software only allows you to create flat keyword lists. A flat keyword list means that there is no logical relationship between keywords. In addition a flat system requires that every keyword for a given picture be entered separately; that is the program won’t apply more keywords based on how the ones applied have been grouped.

More advanced image management programs now usually support keyword hierarchies. That is you can group specific keywords in more general categories and the software will automatically apply the categories as keywords when you apply the specific one.

For example, say you have a keyword hierarchy setup as shown above. When the keyword leopard is applied to an image, the library software will also recognize that the image should also have the keywords “cats” and “animals”. In a flat keywording system all three keywords must be applied manually.

The advantage of this is obvious, it becomes much easier and straight forward to apply rich descriptive sets of keywords. Besides, the obvious improvements in the speed you can keyword things, it also helps make things more consistent. A picture of a lion only need be tagged lion, remembering if it also should have “cats” or “cat” or “big cat” becomes something the software takes care of.

Creating Logical Hierarchies

The one problem that a hierarchical keywording system poses is creating the logical hierarchy. Again it’s important to understand that there is no right and wrong way to go about doing this, and a lot of how things are organized depends on the person doing the organization. Remember, this is an aid for finding pictures, rigidly adhering to a strategy that makes no sense to you won’t help.

Above are two examples of how one might organize keywords relating to a certain African big cat. In figure 3, the choice was made to intermix the sex and species, that is the keyword lion specifically refers to a male “lion” and the keyword “lioness” refers to a female lion.  Figure 4 shows an alternative approach, in this case the sex and species have been separated, a picture of a female lion would carry the keywords “lion” and “female”.

There are pros and cons to each way, and these aren’t the only possible options. The idea though is to develop a strategy that adheres to a consistent convention. If you’re primarily an African wildlife photographer and you expect people to logically look for your images of female lions by searching for “lioness” then the first system would likely be more efficient both for entry and for generating search hits. Conversely if the bulk of your work revolved around species that generally are difficult to differentiate visually, the sex isn’t important or there isn’t a separate word for the male and females, the second solution may work better.

There are other strategies as well. Both systems could be combined, resulting in a picture of a lioness being tagged with “lioness” and “female”, or even something completely different could be used. The ultimate goal though is to devise a system that works for you and takes advantage of natural groupings to add more information to the image.

Getting your Keywords into your Library Software

There are generally three ways to go about getting the actual keywords into your software so you can use them. While the details vary, the broad methods are importing a pre-built keyword list, creating them manually in the program’s keyword list, or creating them as you go by typing them into the keyword entry textbox. Each method has advantages and disadvantages.

Creating your keywords though importing a keyword file is the fastest way to build large collections of keywords. Typically, the keyword files are simple text files with some form of markup to indicate their relationship and they should treated.

The second most useful method is to add keywords though the keyword entry text box. This is you can type keywords as you go though your imported pictures. This offers less flexibility than the adding them through the keyword list but tends to be a bit quicker when you’re working though a set of images. Generally there is a way to indicate that a keyword typed should be considered a child of another keyword.

The last method is to create the keywords manually though the keyword list. How exactly this is done, is program specific but generally is done some control on or in the keyword list. The resulting keyword dialog box typically offers a way to add synonyms and tell the software whether it should export the keyword or not, among other options.

Next time I’ll cover creating keyword files to import into Adobe’s Photoshop Lightroom.