We regularly get queries/comments from community members about the algorithm for MediaSearch

We currently have this page explaining how it works, but it needs to be updated with the latest developments (weighted_tags and synonyms) and probably could do with a simplified intro to go with the more detailed explanation below e.g. "MediaSearch looks for your search term in title, category, description, .... how much weight is given to each field is based on a model optimising for highest number of 'good' images in the first 25 results from a training dataset of 12k rated image results"