Learning to rank

Learning to rank or machine-learned ranking (MLR) is the application of machine learning, typically supervised, semi-supervised or reinforcement learning, in the construction of ranking models for information retrieval systems

Resources

• https://en.wikipedia.org/wiki/Learning_to_rank
• Training data consists of lists of items with some partial order specified between items in each list. This order is typically induced by giving a numerical or ordinal score or a binary judgment (e.g. "relevant" or "not relevant") for each item. The ranking model's purpose is to rank, i.e. produce a permutation of items in new, unseen lists in a way which is "similar" to rankings in the training data in some sense.

Ordinal regression (classification)

• https://en.wikipedia.org/wiki/Ordinal_regression
• OR (also called "ordinal classification" or “ranking learning”) is a type of AI/Supervised Learning/Regression analysis used for predicting an ordinal variable, i.e. a variable whose value exists on an arbitrary scale where only the relative ordering between different values is significant. It can be considered an intermediate problem between regression and classification.
• http://stackoverflow.com/questions/3495157/ordinal-classification-packages-and-algorithms
• http://stats.idre.ucla.edu/r/dae/ordinal-logistic-regression/

Code

• #CODE RAX
  • #PAPER Rax: Composable Learning-to-Rank using JAX (Jagerman 2022)
  • https://www.marktechpost.com/2022/08/16/google-ai-open-sources-rax-a-python-library-for-ltr-learning-to-rank-in-the-jax-ecosystem/
• #CODE Adarank
• #CODE Pyltr - LambdaMART
• #CODE Mord - Ordinal Regression in Python
  • https://pythonhosted.org/mord/
  • http://fa.bianp.net/blog/2013/logistic-ordinal-regression/

References

DL-based ranking

• #PAPER TF-Ranking: Scalable TensorFlow Library for Learning-to-Rank (Kumar Pasumarthi 2019)
  • #CODE https://github.com/tensorflow/ranking
  • New Keras-based TF-Ranking version