Aprimorando a instalação e a configuração de experimentos do RecSysExp. Silva, S. C. d. Technical Report Universidade Federal de Ouro Preto, Ouro Preto, BR, 2024. Accepted: 2024-02-29T14:36:20Z
Aprimorando a instalação e a configuração de experimentos do RecSysExp. [link]Paper  abstract   bibtex   
The paper presents significant enhancements to the RecSysExp framework, used for conducting experiments in recommendation systems. These improvements were aimed at enhancing the usability, scalability, and readability of the system. The new functionalities cover three distinct areas: the development of a graphical user interface, the encapsulation of the framework using Docker, and the restructuring of a class for more cohesive integration with datasets, following established design patterns. The primary goal was to enhance the value provided by the framework, aligned with the vision of its creators, aiming at its use as an academic tool in classroom or research environments. The methodological approach adopted employed specific technologies for each addressed context. For the creation of the user interface, React and Next.js frontend frameworks were employed, while Dockerfile and docker-compose were used for the encapsulation of RecSysExp. Finally, the modification of the class responsible for datasets was carried out following the Template Method design pattern. The project successfully achieved all proposed objectives. The implementation of a container structure simplified the installation of the system, while improvements in the visualization of configurations made experiment creation more intuitive. Additionally, the ability to upload files expanded user options. Although the final version of RecSysExp functions similarly to its original iteration, the additions from this work resulted in an enhanced and more user-friendly version. However, it is important to note that configuration through the graphical interface has limitations, as it is only possible to configure algorithms and modules that can be instantiated via configuration files in the framework. Algorithms and modules implemented solely as libraries in other projects cannot be configured via the frontend.
@techreport{silva_aprimorando_2024,
	address = {Ouro Preto, BR},
	type = {Bachelor {Thesis}},
	title = {Aprimorando a instalação e a configuração de experimentos do {RecSysExp}.},
	url = {http://www.monografias.ufop.br/handle/35400000/6571},
	abstract = {The paper presents significant enhancements to the RecSysExp framework, used for conducting experiments in recommendation systems. These improvements were aimed at enhancing the usability, scalability, and readability of the system. The new functionalities cover three distinct areas: the development of a graphical user interface, the encapsulation of the framework using Docker, and the restructuring of a class for more cohesive integration with datasets, following established design patterns. The primary goal was to enhance the value provided by the framework, aligned with the vision of its creators, aiming at its use as an academic tool in classroom or research environments. The methodological approach adopted employed specific technologies for each addressed context. For the creation of the user interface, React and Next.js frontend frameworks were employed, while Dockerfile and docker-compose were used for the encapsulation of RecSysExp. Finally, the modification of the class responsible for datasets was carried out following the Template Method design pattern. The project successfully achieved all proposed objectives. The implementation of a container structure simplified the installation of the system, while improvements in the visualization of configurations made experiment creation more intuitive. Additionally, the ability to upload files expanded user options. Although the final version of RecSysExp functions similarly to its original iteration, the additions from this work resulted in an enhanced and more user-friendly version. However, it is important to note that configuration through the graphical interface has limitations, as it is only possible to configure algorithms and modules that can be instantiated via configuration files in the framework. Algorithms and modules implemented solely as libraries in other projects cannot be configured via the frontend.},
	language = {pt\_BR},
	urldate = {2024-10-11},
	institution = {Universidade Federal de Ouro Preto},
	author = {Silva, San Cunha da},
	year = {2024},
	note = {Accepted: 2024-02-29T14:36:20Z},
}

Downloads: 0