User-Item Predicted Relevance Function

A User-Item Predicted Relevance Function is a item relevance measure that is a ...

• AKA: pItemRelevance().
• Context:
  • It can range from being a Personalized pCTR() [math]\displaystyle{ p(click \mid item, context) }[/math] to being an Non-Personalized pCTR().
  • It can range from being a Contextual pCTR() [math]\displaystyle{ p(click \mid item, user) }[/math] to being an Non-Contextual pCTR().
  • It can be produced by a Clickthrough Rate Prediction System (that solves a clickthrough rate predictive modeling task).
  • …
• Example(s):
  • a Clickthrough Rate Estimation Function, when used as a relevance function.
  • …
• Counter-Example(s):
  • a Historical Clickthrough Rate.
  • an eEPC() Estimation Function/Predicted Earnings-per-Click Model/pEPC().
  • a Item Rating Prediction Model (for item ratings).
  • a Dwell Time Prediction Model.
• See: Consumer Experience Prediction Model.

---

---