Modeling click-through rates in social networks with martingales

The celebrated martingale convergence theorem states that bounded martingales always converge to a limit with probability one. Less is known about the rate of convergence of such martingales towards their limit. Previously, we analyze the rate of convergence of various bounded martingales [1]. In this work, click-through rates in social networks will be modeled by bounded martingales to quantify the success of an online advertising campain for a particular website.

Supervised by Nezihe Merve Gürel

nezihe.guerel@inf.ethz.ch

References

  1. [1] Gurel, N. M., & Leveque, O. (2015). On the convergence of some bounded martingales.
  2. [2] Chen, K., Cai, K., Huang, L. & Lui, J. C. S. (2018). Beyond the Click-Through Rate: Web Link Selection with Multi-level Feedback and references therein, arXiv: 1805.01702.