Publications

USEFUL
MendeleyMIT Labraries Overleaf PhD Student Start Zotero

RESEARCH MANIFESTOS OF POSTGRADUATE STUDENTS OF THE TEAM

Use of Machine Learning in Software Engineering to Build a Framework Aiming to Support the Application of Compliance in the Audit and Control of Public Services
Álvaro Farias Pinheiro, PhD Student – MSc
www.alvarofpinheiro.eti.br

My current research work will lead to my Ph.D. in Computer Engineering at the University of Pernambuco, Brazil. It envolves the integration of intelligent techniques and a specially tailored methodology for development, validation, and varification of compliance in the public service. Should my work strikes success, regiments, rules, standards, laws, policies, and regulations of public systems will guide auditing, control systems in a more compliance wanner. The end result will include a framework, aiming to subsidize the development of intelligent tools for the enforncement of preventive and repressive measures against corruption. We hope that by using Artificial Intelligence and Data Mining associated with the future proposed framework and methodology, information and decision making to curb corruption will exihibit performance and qualitative gains. To achieve that, my plan is to initially perform a systematic literature review on current methods and technologies, especially focusing on software engineering that applies AI as the core processing approach towards compliance. Some examples of possible soluctions to conflicts that often exist in relation to conformity in public systems. These conflicts, are likely to be due to the gigantic data volume and inherently complexity. Either one strong justification for AI utilization.


ATTORNEY GENERAL FROM THE STATE OF PERNAMBUCO
SYSTEM TEAM
GOVERNMENT INNOVATION LABORATORY
ACTIVITY DESCRIPTION START PHASE
Active Debt Rating Application of Artificial Intelligence, able to classify active debts, from good to not so good, to be collected in order to increase revenue, and identify relevant points not observed or ignored by experts. May 15, 2019 Homologation
Robot Extractor Robot Process Automation (RPA) designed to automate the activity of extracting extracts from debtors and legal entities. June 21, 2019 Production
Miner System developed for the purpose of automatic extraction of judicial acts for correction. August 1, 2019 Production
Robot Search Personal Property Robot Process Automation (RPA) developed for the purpose of automating the activity of extracting the extracts from assets of debtors and legal entities. July 12, 2019 Homologation
Dashboards Graphical panels that make it possible to analyze data and make independent data discoveries. Being able to share knowledge and data analysis in groups and among organizations. August 5, 2019 Development


Dissertation topic
Caio Melquiades, Master’s Student – BSc
CV (Lattes Platform)

My research’s goal is to develop a compliance aware (e.g. conformity to regulations) solution to search for evidence of non-compliant past actions taken from public agents or private institutions and individuals in monetary relations with public service in large, multi-source and multidimensional databases. This kind of search in past (already finished) actions has as one of it’s main features the so-called batch learning, that tries to maximize accuracy, in contrast to on-line classification, that learns from every new piece of data, in order to update his assessment ability all the time. Due to the constitutional principles related to the transparency of public administration, this auditing solution shall be also capable of explaining what factors were taken into account when the classification and / or search actions are done, avoiding the black box problem and biased decisions that computational intelligence algorithms sometimes accidentally make. As a validation tool, an auditing application will be developed to use this combination of concepts and techniques to find corruption evidences in large public databases.



Dissertation topic
Renato Barbosa Cirne, Master’s Student – BSc
CV (Lattes Platform) :: Research Project Abstract

My research aims to develop a method of selecting characteristics in the identification of error or fraud in business processes using Artificial intelligence techniques. It is expected that the usage of Artificial Intelligence techniques could identify a variety of probable fraud pattern practices and corruption manners, based on prior knowledge, as well as to support the development of predictive models for the assessment of risk areas. To adapt compliance programs and contribution to a safer business environment, especially in the public sector, is also highly regarded.



Dissertation topic
João Alberto da Silva Amaral, Master’s Student – BSc
CV (Lattes Platform)

My research aims to develop a model using techniques of artificial intelligence that helps the internal control system to classify the purchases of goods and services carried out by the state according to the risks of which they incur in damages to the state treasury. In order to do so, the data of the punishment records applied to the suppliers together with the data of the respective bidding processes will be analyzed, catering for a suitable identification of nonconformities related to each type of punishment. In case of success, it will be possible to increase the activities of the three levels of Control because we will be able to classify future purchases in advance. Thant for proactivelly intervene and remedy inconsistencies, as well as to act in a corrective way in the follow-up of processes that present a greater risk of damages to the public administration. Another possible gain is in the performance of internal control and alignment with the strategic planning of the state institutions.