Presentation information

General Session

General Session » [GS] J-2 Machine learning

[4O2-J-2] Machine learning: improvements of user satisfaction

Fri. Jun 7, 2019 12:00 PM - 1:20 PM Room O (Front-left room of 1F Exhibition hall)

Chair:Yoshifumi Seki Reviewer:Hidekazu Oiwa

12:00 PM - 12:20 PM

[4O2-J-2-01] Deep Reinforcement Learning for Recommender System of Users’ Behavior on Website

〇Kazuma Minoda1, Hiromu Auchi1, Nobuyuki Kawagashira2, Nobuyuki Ishikawa1 (1. Recruit Technologies Ltd., 2. Private Author)

Keywords:Deep Reinforcement Learning, recommendation

For the companies running websites,we need appropriate communication to the visiting users according to their situation such as pages they have visited so far.To achieve that purpose,we should have some strategies to suggest the users to take desirable actions on the website. As one of the strategies,recommendation of users’ behavior can be considered.When we recommend some actions not directly related from conversion pages (e.g.reservation pages,purchase pages,etc.) on website such as selecting the searching conditions,it is impossible to recommend the best action that leads to the user's reservation behavior by using the conventional supervised learning methods.In this research,we solve the above problem using deep reinforcement learning method and show its effectiveness by an experiment for actual web access log of users’ behavior.