Word embeddings in electronic invoices to support fraud detection

• Grant number: 22/09232-0
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: September 01, 2022
• End date: June 30, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: José Alberto Cuminato
• Grantee: Carllos Eduardo Alves de Holanda
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 13/07375-0 - CeMEAI - Center for Mathematical Sciences Applied to Industry, AP.CEPID

Abstract

Currently, there are no concrete studies of the textual description of electronic invoices for the extraction of relevant information to be used, for example, in the detection of product overprices and overpricing of public contracts. In general, electronic invoices contain insufficient and unstructured information causing great difficulty in determining products and services. In order to assist in the systematic detection of fraud in public and private sectors, this project aims to study representations of data from electronic invoices via word embeddings. In addition to the use of some well known algorithms like word2vec, seeking greater efficiency of representations, we will also consider the study of embeddings in non-Euclidean spaces. In particular, a study of representation and visualization of invoice data in hyperbolic spaces will be considered. (AU)