Learning to Rank

Student seminar — Session 22

Focus¶

Learning to rank: problem settings (pointwise, pairwise, listwise), evaluation metrics (Precision@k, MAP, NDCG), and representative methods.

Reading (examples)¶

• Zaki & Meira, ranking / evaluation chapters

• Liu, Learning to Rank for Information Retrieval (selected sections)

• A recent survey or application paper (to be assigned)

Student tasks¶

• Explain the problem formalization and assumptions

• Present one method and its objective/loss

• Discuss evaluation protocol and limitations

• Q&A and short discussion