Tourist Profiler

This work was supported by the research project RASTA: Realtá Aumentata e Story-Telling Automatizzato per la valorizzazione di Beni Culturali ed Itinerari; funded by the Italian Ministry of University and Research (MUR) under the PON Project ARS01_00540.

Dataset

The dataset folder contains the methods used for creating the dataset utilized in training and testing the models. Additionally, it includes a notebook for analyzing the dataset, allowing for a detailed exploration of its characteristics and insights.

DQN

The dqn folder contains the implementation of the DQN (Deep Q-Network) model. It includes the following files:

• Agent_DQN.py: Contains the implementation of the DQN agent.
• Environment_DQN.py: Contains the implementation of the environment used for the model.
• Dataset_DQN.py: Implements methods for managing the dataset.
• Settings_DQN.py: Includes the configuration settings for the DQN model.

To run the experiment, use the notebook dqn.ipynb, which provides a complete process for training and testing the DQN model.

Visualization

Below is an image representation of the DQN model:

HRL

The hrl folder contains the implementation of the HRL (Hierarchical Reinforcement learning) model, combined with a Matrix Factorization model designed to infer the missing values in the Profile. It includes the following files:

• Agent.py: Contains the implementation of the HRL agent.
• Environment.py: Contains the implementation of the environment used for the model.
• Recommender.py: Contains the implementation of the NAIS recommender model used combined with HRL agent.
• DataMovielens.py: Implements methods for managing the dataset.
• Settings.py: Includes the configuration settings for the HRL model
• Evaluation.py: Implements the calculation of some evaluation metrics for the model
• Utils.py: Implements some functions for training and evaluating the model

Visualization

Below is an image representation of the HRL model:

Profile Models

The profile_models folder contains the implementation of the baseline profile models used for comparison against our model. This folder includes all the necessary files to run the experiments, along with a notebook that provides a complete workflow for training and testing these models.