saltyyyyyyy/thompson-sampling-for-a-fatigue-aware-online-recommendation-system

Thompson-Sampling-for-a-Fatigue-aware-Online-Recommendation-System

Introduction

This python code is for paper ''Thompson Sampling for a Fatigue-aware Online Recommendation System''. Please find details on https://arxiv.org/abs/1901.07734

This code implements algorithm 1, 2, 3. It is used for performing the experiments in Section 7.

The result file and all the charts are the result when we run the code.

• Based_define_function.py: This code provide the basic function used in the paper, i.e. calculate the order of messages, the users feedback, the total payoff, etc.

• Thompsonbasedpy.py: This code provides function thompson_base that we implement algorithm 1.

• TSgaussian.py: This code provides function TSgaussian_base that we implement algorithm 1.

• UCBV.py: This code provides function ucbv that we implement algorithm 3.

• ucbbasedpy.py: This code provides function ucb_base that we implement algorithm in Cao and Sun (2019).

• experiment.py: This code calls the function from other py file. The current experiment.py file is that we call algorithm 2 to do 10 times experiments using different R alpha and beta. We record the result in txt file and pickle down in result file.

• testfile.py: This code plots the chart of the result of each algorithm in the result file. The current testfile.py file is that we plot the result chart of algorithm 1 when u is generated from [0, 0.5].