Due to the competitive nature of e-Commerce and proprietary information and tools used and created in the majority of my work at Getty Images, deliverables demonstrating my learning objectives will be summaries and broad descriptions.

In order to meet this objective, I worked mainly in the Editorial content of Getty Images photographs and footage. Within the Editorial content, Preferred Terms of people, places, and events are added to aid in retrieval of the photographs and footage. Many of the names of people are ambiguous, as many people in different industries share the same name. For example, Stephen King could be an author, a horse jockey, a rugby player, or an NFL player. In order to map the appropriate photographs or footage to the correct Stephen King, at least one, if not more, disambiguation rule using Boolean search strings needs to be created. In order to write disambiguation rules, a solid understanding of the structure of the other preferred terms in the controlled vocabulary are applied to photographs and footage. For example, Stephen King the NFL player would have football terms and the team he plays for connected to it. These other terms are helpful in writing disambiguation rules. Over the course of the quarter, I wrote 195 disambiguation rules.

In order to assist in user search and asset retrieval, Getty Images has a very large controlled vocabulary with thousands of keywords and preferred terms for people, events, and locations. Through all of my projects at Getty, I worked with the controlled vocabulary, both directly and indirectly. Directly, I added preferred terms and suggested changes for the structure of keywords as appropriate. Indirectly, I had to have a solid understanding the hierarchy of the controlled vocabulary in order to write disambiguation rules, as well as work with keywords associated with images and footage to implement changes to search results in terms of relevancy.

Within search results on a consumer facing eCommerce site, there are many ways to display the results. Search results may come up due to recency, relevancy, or those that are "best sellers." In order to enhance search results and image and footage retrieval for Getty's customers, I worked on a project that looked at keywords and how they are applied to images and footage to return the most precise and accurate, or relevant, search results. Through this project, I examined top keyword searches, and made changes internally to ensure the most "iconic," or relevant, images and footage floated to the top of the search results. This helped me to understand the intricacies within keywording, as well as working with millions of images and footage. Even within relevancy, there is a degree of ambiguity, as you don't always know exactly what your customer wants when searching on a keyword. For example, "glasses," as in glasses you wear to help you see, is unclear. When a customer searches on "glasses," are they looking for just glasses as objects by themselves? Or are they looking for people wearing glasses? Are they looking for just regular glasses or sunglasses as well? This is just an example of the multitude of questions I had to ask when working with keywords to produce the most relevant results, as relevancy can often be very subjective.