Sampling Bias Due to Near-Duplicates in Learning to Rank.
Maik FröbeJanek BevendorffJan Heinrich ReimerMartin PotthastMatthias HagenPublished in: SIGIR (2020)
Keyphrases
- learning to rank
- active sampling
- ranking functions
- information retrieval
- loss function
- evaluation measures
- direct optimization
- ranking svm
- document retrieval
- collaborative filtering
- random sampling
- evaluation metrics
- retrieval systems
- query dependent
- exploration exploitation dilemma
- test collection
- sample size
- supervised learning
- active learning
- ranking models
- query expansion
- learning to rank algorithms
- support vector