Inferring about fraudulent collusion risk on Brazilian public works contracts in official texts using a Bi-LSTM approach

Findings of the Association for Computational Linguistics: EMNLP 2020, pages 1580–1588 November 16 - 20, 2020. c⃝2020 Association for Computational Linguistics 1580 Inferring about fraudulent collusion risk on Brazilian public works contracts in official texts using a Bi-LSTM approach Marcos Cavalcanti Lima Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil marcosc.mcl@pf.gov.br Leonardo R. de Carvalho Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil leouesb@gmail.com Felipe L. de Souza Mendes Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil felipelopess@gmail.com Roberta da Costa Silva Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil robertacs42@gmail.com Aleteia Araujo Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil aleteia@unb.br Fl´avio de Barros Vidal Dept. of Computer Science University of Bras´ılia (UnB) Bras´ılia - DF - Brazil fbvidal@unb.br Abstract Public works procurements move US$ 10 bil- lion yearly in Brazil and are a preferred field for collusion and fraud. Federal Police and au- dit agencies investigate collusion (bid-rigging), over-pricing, and delivery fraud in this field and efforts have been employed to early de- tect fraud and collusion on public works pro- curements. The current automatic methods of fraud detection use structured data to classi- fication and usually do not involve annotated data. The use of NLP for this kind of ap- plication is rare. Our work introduces a new dataset formed by public procurement calls available on Brazilian official journal (Di´ario Oficial da Uni˜ao), using by 15,132,968 textual entries of which 1,907 are annotated risky en- tries. Both bottleneck deep neural network and Bi-LSTM shown competitive compared with classical classifiers and achieved better preci- sion (93.0% and 92.4%, respectively), which signs improvements in a criminal fraud inves- tigation. 1 Introduction In the last five years, Brazil’s federal government invested (Brazil, 2020b) in 23,352 public works contracts adding up to R$ 283.8 billion (approx. US$ 49.3 billion in May 2020). Those works con- sist of all sorts of projects from oil refineries to ports, from soccer stadiums to power plants, from tunnels to dams, and are developed on a continental- sized territory, generating an endless and growing quantity of information regarding those projects. Thereupon, public works procurements are a pre- ferred field for collusion and fraud (OCDE, 2007). Brazilian Federal Police have been working on fraud investigations on public works for the last four decades and develop its investigation based on a highly skilled group of experts formed by civil, electrical, mechanical, computer engineers, and accountants (APCF, 2020). The types of fraud in- vestigated are mainly collusion (bid-rigging), over- pricing, and delivery fraud (quality and quantity of services and materials). We will bring the knowl- edge accrued during those decades to enhance our data understanding (Lopes, 2011). As described in Foundation (2020), public works contracts are made via procurement, that is the pro- cess of public administration uses to make all its contracts. Every procurement step is usually publi- cized by a call for …

Content truncated to 500 words