AUTO-DataGenCARS+: Advanced Python-Based User orienTed tOol DataGenCARS

Brief description

AUTO-DataGenCARS+ is a powerful graphical user interface that can be used to generate synthetic data for the evaluation of Recommender Systems (RS) and Context-Aware Recommender Systems (CARS). It extends our previous tool DataGenCARS with a flexible and useful GUI, user facilities, improved and new functionalities like the following:

• Generate a synthetic dataset (with explicit and implicit ratings).
• Pre-process a dataset:
• Generate and replace null values in item and context files.
• Automatically generate user profiles.
• Transform attributes:
• Convert data between numerical and categorical representations.
• Transform ratings based on value ranges (e.g., [1-5]) into binary values.
• Analyze a dataset:
• Visualize and analyze datasets (with explicit and implicit ratings).
• Evaluate recommendation systems (traditional RS and CARS) using experimental settings.

IMPORTANT: this webpage focuses on AUTO-DataGenCARS+, a more advanced and new version of the tool AUTO-DataGenCARS, now developed in Python. Originally, AUTO-DataGenCARS was implemented using Java. The AUTO-DataGenCARS invention has been registered in Spain (University of Zaragoza — PII-2021-0019).

Documentation

AUTO-DataGenCARS+ provides 'Help Information' for each functionality, which includes some videos demonstrating example scenarios of user interaction.

Software

• Source code for AUTO-DataGenCARS+ (public)

---

• Demo instance of AUTO-DataGenCARS+ (credentials could be requested from us)
• Private source code for AUTO-DataGenCARS+
• Private GitLab repository for AUTO-DataGenCARS+ (Python version)
• Private GitHub repository for AUTO-DataGenCARS (Java version)
• Private GitHub repository for the initial version of AUTO-DataGenCARS (Java version)

Videos

• Analysis of a dataset/Visualization: RS dataset with explicit and preferential ratings
• Analysis of a dataset/Visualization: CARS dataset with explicit and preferential ratings
• Analysis of a dataset/Evaluation: RS with explicit and preferential ratings
• Analysis of a dataset/Evaluation: CARS (CM paradigm) with explicit and preferential ratings
• Generate a synthetic dataset: CARS dataset with explicit and preferential ratings
• Generate a synthetic dataset: RS dataset with explicit and preferential ratings
• Generate NULL values in an item file: Traditional RS
• Generate NULL values in an item file: CARS
• Generate NULL values in a context file: CARS
• Replace NULL values in an item file: RS
• Replace NULL values in an item file: CARS
• Replace NULL values in a context file: CARS
• Generate a user profile: CARS
• Generate a user profile: RS
• Use Case: Generate a Synthetic Dataset
• Use Case: Enlarge an Existing Dataset
• Use Case: Incorporate Context Data into an Existing Dataset
• Use Case: Reduce Bias in an Existing Dataset

Contributors

• Sergio Ilarri
• María del Carmen Rodríguez-Hernández
• Raquel Trillo-Lado
• Ramón Hermoso

• Marcos Caballero Yus (final master’s project)

AUTO-DataGenCARS+ vs. DataGenCARS

A comparison between AUTO-DataGenCARS+ and DataGenCARS can be seen in the following table:

Acknowledgments

• Project PID2020-113037RB-I00 (funded by MICIU/AEI/10.13039/501100011033) — Next-gEnerATion dAta Management to foster suitable Behaviors and the resilience of cItizens against modErN ChallEnges (NEAT-AMBIENCE).
• Government of Aragon (COSMOS research group; last group reference: T64_23R; previous group reference: T64_20R).
• This work has also been partially funded by the Department of Big Data and Cognitive Systems at the Technological Institute of Aragon, by DGA-FSE IODIDE research group of the Government of Aragon (grant number T17_23R) and by the European Regional Development Fund (ERDF).

Logos