A model developed in Brazil can predict a convict
image: A study by researchers at the University of São Paulo estimated the likelihood of a future conviction of politicians for corruption and other financial crimes by analyzing networks pointing to the similarity of voting histories.
see Following
Credit: Laycer Tomaz / Câmara dos Deputados
Birds of a feather fly together. The popular saying applies to politics and computational analysis of complex networks in corruption research, judging by a study by scientists at the University of São Paulo in Brazil.
According to the authors, it is possible to predict whether MPs (members of Câmara dos Deputados, the lower house of Congress in Brazil) will be convicted in the future for corruption or white-collar crime by analyzing the similarity between their voting results. and those of legislators already condemned.
The study is published as a chapter of the book Corruption networks. Researchers analyzed the voting and sounding histories of 2,455 politicians elected to the lower house between 1991 and 2019, over 3,407 sessions in total involving votes on bills covering a wide range of topics.
âThe surprising aspect of the study is that we did not need to use data from court cases to find this correlation between voting history and corruption. We have only used home vote logs to create networks showing what we call the âvoting neighborhoodâ in terms of how and with whom MPs voted. Based on this, our model can predict whether an MP is corrupted with 90% accuracy, “said Tiago Colliri, co-author of the study.
The analysis was carried out during Colliri’s doctoral research at the São Carlos Computer Science Department of USP, with a grant from the National Council for Scientific and Technological Development (CNPq). FAPESP provided support through a Thematic project and the Artificial intelligence center (C4AI), a partnership between FAPESP and IBM.
The analysis of complex networks has been applied very widely, including in fields such as biological neural networks and food chains, for example. In crime-related studies, goals ranged from finding a correlation between social capital and corruption risk in local government contracts to identifying hidden links between members of an Italian mafia group.
To understand the approach, it is necessary to keep in mind that complex networks refer to large-scale graphs with non-trivial connection models. âOne of the main characteristics of any complex network is the modeling of various types of relationships between nodes – or assistants, in the case of our study. These can be local, intermediate or global relationships. Here we set out to identify the relationship between MPs and how they voted in Congress, and our method was very accurate for prediction purposes, âsaid Zhao Liang, professor in the Department of Computer Science and Mathematics of the Ribeirão Preto School of Philosophy, Sciences and Letters of USP (FFCLRP) and the other co-author of the study.
Similarity
After creating a network based on the voting histories of nearly 2,500 MPs, the researchers noted that some MPs who had been notoriously convicted of corruption had voting histories similar to those of other MPs. “In this type of network analysis, each node represents an MP and each edge represents the similarity of votes between a pair of MPs,” Colliri explained.
The researchers detected a pattern in the network. âThe diagram showed a proximity of vote or a similarity between the deputies whose convictions had been reported in the media. There was a consonance in their voting stories, âhe said.
To validate the findings, they assembled a separate database with data on MPs convicted of corruption from sources such as the Supreme Court of Brazil (STF). âWith this secondary database, we checked 33 people who had been convicted, and they were not dispersed but clustered in the network,â Colliri said. âThey have formed a model of what we call ‘neighbors of corruption’. We then tested this map using a number of link prediction algorithms based on common neighbors. The algorithms have been shown to be able to predict whether an MP is corrupted with 90% accuracy. “
One of the conclusions to be drawn from the study is that corruption in Congress can be controlled in an easier way. âWe have found that corrupt MPs vote the same in our Congress, so that a predictive model can be obtained more simply and control can be much easier to do. It is much easier to analyze this data than it is to search through lawsuits, criminal trials, media reports and even family trees, âZhao said.
About the São Paulo Research Foundation (FAPESP)
The São Paulo Research Foundation (FAPESP) is a public institution whose mission is to support scientific research in all fields of knowledge by awarding scholarships, scholarships and grants to researchers linked to educational institutions Higher and Research Institute of the State of São Paulo, Brazil. FAPESP is aware that the best research can only be done by working with the best researchers at the international level. Therefore, it has established partnerships with funding bodies, higher education institutions, private companies and research organizations from other countries known for the quality of their research and has encouraged scientists funded by its grants to further develop their international collaboration. You can find out more about FAPESP at www.fapesp.br/fr and visit the FAPESP press agency at www.agencia.fapesp.br/fr to keep abreast of the latest scientific advances, FAPESP contributes to the realization of its numerous programs, prizes and research centers. You can also subscribe to the FAPESP press agency at http://agencia.fapesp.br/subscribe.
The title of the article
Predict Corruption Convictions Among Brazilian Representatives Using Voting History-Based Network
Publication date of the article
25-Sep-2021
Warning: AAAS and EurekAlert! are not responsible for the accuracy of any press releases posted on EurekAlert! by contributing institutions or for the use of any information via the EurekAlert system.