learning to rank

• These are problems where you tell a computer to rank a list of n items (based on relevance)
• You want a model that scores items (like pagerank), so you can just sort each item by their score
• three approaches:
  • https://notesonai.com/Learning+to+Rank
  • Pointwise Approach
    • Predict the relevance per item, simple but very naive.
    • Ignores that ordering of items is what matters.
  • Pairwise Approach
    • Loss based on document pairs, minimize the number of incorrect inversions.
    • Ignores that not all document pairs have the same impact.
    • Often used in the industry.
  • Listwise Approach
    • Tries to optimize for IR metrics, but they are not differentiable.
      • what is an IR matrix?
    • Approximations by heuristics, bounding or probabilistic approaches to ranking.
• Check out these ranking loss functions
• Do not confuse learning to rank with ordering objects in list (doesn’t have a query vector)