Deep Reinforcement Learning for Online Advertising in Recommender Systems

  • 2019-09-10 02:54:37
  • Xiangyu Zhao, Changsheng Gu, Haoshenglun Zhang, Xiaobing Liu, Xiwang Yang, Jiliang Tang
  • 0

Abstract

With the recent prevalence of Reinforcement Learning (RL), there have beentremendous interests in utilizing RL for online advertising in recommendationplatforms (e.g. e-commerce and news feed sites). However, most RL-basedadvertising algorithms focus on solely optimizing the revenue of ads whileignoring possible negative influence of ads on user experience of recommendeditems (products, articles and videos). Developing an optimal advertisingalgorithm in recommendations faces immense challenges because interpolating adsimproperly or too frequently may decrease user experience, while interpolatingfewer ads will reduce the advertising revenue. Thus, in this paper, we proposea novel advertising strategy for the rec/ads trade-off. To be specific, wedevelop a reinforcement learning based framework that can continuously updateits advertising strategies and maximize reward in the long run. Given arecommendation list, we design a novel Deep Q-network architecture that candetermine three internally related tasks jointly, i.e., (i) whether tointerpolate an ad or not in the recommendation list, and if yes, (ii) theoptimal ad and (iii) the optimal location to interpolate. The experimentalresults based on real-world data demonstrate the effectiveness of the proposedframework.

 

Quick Read (beta)

loading the full paper ...