What is TF-IDF Search Relevance Model?
TF-IDF (term frequency weighted by inverse document frequency) is a relevance model that determines how relevant a particular document is for a given query by weighting the number of times that query appears within the searchable text corpus (TF) by the number of times that query appears within the specific document (IDF). TF-IDF serves as the foundation for determining relevance in many open source search engine software packages. The reason this model is so prevalent is because it assigns a relevance score for each document that results can then be sorted upon, as opposed to a simple yes/no relevance model which merely determines if a document is relevant or not for that query and returns disorganized results with no relevance score.
Swiftype search engines use TF-IDF as a starting point for determining relevance, but are also built to learn from user clickthrough behavior. Furthermore, site owners can use the Weights tab in their Swiftype dashboard to precisely control how certain elements of their website schema impact search results.