Ranking Evaluation API Inquiry

After reading about Ranking Evaluation API, and I am planning to use it in my search engine project, however, the following points were not clear to me

  1. Does it enhance search results as per query per user (set of users)? Enhances the overall searching performance?
  2. Can I add semantic search to it in the search query? for example, vectorizing the user's query & documents -in the indices- via pretrained Language model and compare them via CosineSimilarity?
  3. After reading about it, I understand that it might be like a big table with a popularity score?
1 Like