A new pointwise learn-to-rank algorithm
- Prelegent(ci)
- Mikołaj Fejzer, Piotr Przymus
- Afiliacja
- Uniwersytet Mikołaja Kopernika
- Termin
- 14 maja 2021 14:15
- Informacje na temat wydarzenia
- meet.google.com/jbj-tdsr-aop
- Seminarium
- Research Seminar of the Logic Group: Approximate reasoning in data mining
We’d like to discuss a new pointwise learn to rank algorithm. This algorithm can be applied to problems when training data has binary characteristics (relevant or not relevant to query) but the underlying problem is more granular (ie. relevant and possible semi-relevant documents). The training method is adaptive (does not need manual adjustment).
Proposed approach was already successfully applied to two research problems, to the bug localization problem, and scoping of research papers relevant to survey topics.
Link do spotkania: https://meet.google.com/jbj-tdsr-aop