International Conference on Computer and Knowledge Engineering

Home / 12th International Conference on Computer and Knowledge Engineering

An interactive user groups recommender system based on reinforcement learning

Authors :

Hediyeh Naderi Allaf¹ Mohsen Kahani²

1- Computer Engineering Department, Yazd University, Yazd, Iran 2- Department of Computer Engineering Ferdowsi University of Mashhad Mashhad, Iran

Keywords :

recommender system،interactive recommendation،reinforcement learning

Abstract :

Nowadays, there are countless and diverse user data in various fields, which requires analysis to find a set of target users. To understand and identify users, exploration can be done interactively in several steps. Considering that the goal is to find users and the datasets are not related to each other, so this type of learning causes an exploration error. This issue is far more effective for a recommendation that includes discrete exploration actions. This paper uses semantic similarity techniques between datasets to improve the problem of exploration of user groups. This model was created by using reinforcement learning, which uses a simulated agent to learn a suitable policy for exploration recommendations. In this framework, discovery is an iterative decision-making process that includes various types of discovery actions. An agent represents a set of groups from which target users are selected and then recommends the best action for the next step. The results and experiments show that the agent can learn the policy without gathering previous sessions and finally provide an acceptable recommendation.