search-algorithm-design v1.0.2

Search Algorithm Research

-- basic features

• Suggestions

  • items that are related or similar
  • search resiliancy
    • small user errors shouldn't defer from getting the correct results
    • autocomplete
      • guess user intent based on current entry

• Search Process

  • Ranking
    • One large switch we could control would be ranking. First it would be trivial to add weights to the search and tweak them in the future. These weights could be based on various factors
      • how close the users search is to item
      • how often we sell a product
      • how much margin do we have in a product
    • This information would need to be on a dedicated and contained backend since it would have margin information.

    • this backend would have to scale and would have to be able to handle our server load, which can include 1000 queries a day, with more many more possible for auto complete