top of page

Artificial intelligence in DF tax foreclosure

FAPDF and PGDF evaluate projects to implement efficient judicial collection management of active debt Last Tuesday (9/28), the Federal District Research and Innovation Support Foundation (FAPDF) held, in partnership with the Federal District Attorney General (PGDF), the Final Event of the Technical Workshop of the project “ Artificial Intelligence in Tax Enforcement”. Held at Sebrae Lab, at the Foundation's headquarters building in Brasília, the workshop aimed to present the three projects received under Call 003/2020 of the DF Challenge Program to the Selection Advisory Committee.

The call selected projects that present computational solutions that apply machine learning methods, including deep structured learning, to aid in conducting and solving tax enforcement processes. The aim is to implement efficient management of the Federal DF's Active Debt Collection System.

The opening of the event was attended by authorities and representatives of the two institutions involved in the project. For the FAPDF, the honor device was composed by the Technical-Scientific Coordinator, Juliana Lustosa, and the Superintendent of Science, Technology and Innovation, Renata de Castro Vianna. The PGDF was represented by the president of the Data Science and Artificial Intelligence Working Group, Izabela Frota Melo, and by the chief attorney for Tax Enforcement Actions, Lucas Terto Vieira.

“Tax foreclosure is one of the processes that most delays the Brazilian judiciary and the difficulty in executing these actions creates a feeling of impunity and, consequently, of fiscal injustice. The situation makes it unfeasible, to some extent, even the use of funds in public policies, so I dare say that this project affects the citizen, the taxpayer, the jurisdiction and society”, says Izabela Melo.

The FAPDF superintendent confirmed the importance of the theme, highlighting the project's consequences both for the Federal Court and for the Federal District Judiciary. “It's an old demand from the Judiciary as a whole, as well as from the Attorney's Office, the Federal Justice and the TJDFT. With this project, we are being pioneers in Brasília and FAPDF also supports other initiatives in Artificial Intelligence. The Foundation's desire and mission is to meet the demands presented by the bodies and institutions of the DF, this is the objective of our Desafio DF program. Therefore, in this workshop we hope to find good projects so that we can help solve the great problem of tax foreclosures”, said Renata Vianna.

Finishing the opening speeches, the chief prosecutor for Tax Enforcement Actions pointed out the volume of actions that the PGDF currently accumulates and the need to seek innovative methodologies to improve the execution process. "The DF has a peculiar characteristic because it absorbs municipal and state competences, so we have municipal and state taxes and, therefore, under my leadership, I have 320 thousand tax enforcement actions that run today in the DF, with R$ 30 billion being executed . To do this task I have 30 proxies, so we observe that we will never reach a point where this execution is really effective. So, it is necessary to rationalize, to look for new ways to carry out this execution”, highlighted Lucas Terto Vieira.

According to the Brazilian Institute of Planning and Taxation (IBPT), Brazil has already failed to collect 417 billion a year as a result of tax evasion by companies. The Institute also indicates that undeclared sales by Brazilian companies reach R$ 2.33 trillion a year.

Projects evaluated

The teams responsible for the projects registered in the public call presented their proposals to the Advisory Committee, composed of technical representatives from FAPDF and PGDF.

The first project was presented by the Brazilian Network for Certification, Research and Innovation (RBCIP). The proposal includes a methodology for screening and legal analysis of processes based on Agile Scrum Method associated with a model for projects centered on CRISP-DM data. The project also reviews the use of technological tools such as Optical Character Recognition (OCR), data processing and machine learning. The event was attended by the President of RBCIP, Marcelo Estrela Fiche, and the project coordinator, Ricardo Silva Carvalho.

The second project of the workshop was presented by the Laboratory of Artificial Intelligence at the University of Brasília (AI.Lab/UnB). The team, which includes professionals from the areas of Law, Software Engineering and Artificial Intelligence, proposed a tool capable of extracting data from documents in different formats for the generation of parts, indication of process phases (filing, citation and constriction), realization of acts and generation of petitions to forward the process to the next phase of the tax execution flow.

The last project presented was from the Brazilian Institute of Human, Intelligent, Creative and Sustainable Cities (Ibrachics). The team proposed a system with an architecture based on four

6 views0 comments


bottom of page