Tagging: Making the Most
In order to really make the most of the tagging system, one of our brilliant developers has compiled this helpful list of advice;
- What Constitutes Garbage Tags?
- Some examples" "affordable", "painting", "photograph", "cheap", "premium", "unique" and redundant combinations like "blue cat", "Blue Cat", "cat blue", "cat painting", "unique cat painting", "one of a kind cat painting", "house cat", "sleeping cat"
- Why are the above examples bad?
- They increase what is known as"Term Frequency". If a word like "painting" appears in the tags of every painting in the search index then it does a very bad job of narrowing down search results and is considered less relevant when searching. Similarly the combination example above does very little to narrow down results. It would be better expressed with a few meaningful tags like "cat", "kitchen", "blue" and the rest belong in the description.
- We search inside the description too!
- Making your description descriptive increases uniqueness to your artwork when it comes time to search. Please don't take this as a sign to put single word tags into the description field thinking it will boost uniqueness... Term Frequency is cumulative and doing so would actually work against you. Write a story about motivations for making the piece. Write personal feelings about it. That is the best way to create uniqueness in a description.
- The auto-complete search behaves differently to "see all" out of necessity.
- "Search as you type" is about preempting what someone is typing and quickly returning possibly relevant results. We perform two searches as people type, one to find products that might match what they're typing and the second one for artists that they might be searching for. This search favours speed over accuracy and, as a result, the search results can differ to those you get when clicking "see all". It also searches over less fields
- If a user performs a keyword search (e.g "Forrest painting") the results will always be returned ordered by what is called "relevancy score".
- This score is a decision by the search engine over what it considers to be the most appropriate given the data held on each product.
- What I can say is that the search considers the following fields when making its decision: Product Name, Artist Name, Product Description, Tags, Category Name (e.g painting, relief, print etc), Subject Name (landscapes, nudes, architecture), as well as a few other factors.
To conclude - Because we specifically search category name, subcategory name, style and subject you can see how some of the tags people enter can be seen as redundant ;-)