Reinforcement Learning Based Group Event Invitation Algorithm

Nowadays, we rely increasingly on mobile applications and social networking websites to organize group events and/or search events to participate. Most existing services and platforms focus on distributing event invitations based on user profiles thus neglecting the fact that event attendees can significantly affect each other's degree of satisfaction. To address this issue, we propose a reinforcement learning based group event invitation algorithm that can track relationships of users and use response rate and post-event reviews as feedback to guide invitation receiver selection process. Experimental results indicate that our proposed algorithm achieves better performance in term of satisfaction and connectivity of users when k-core algorithm and greedy search algorithm are compared.