2020
DOI: 10.1007/978-3-030-57672-1_12
|View full text |Cite
|
Sign up to set email alerts
|

Using Machine Learning for Automated Assessment of Misclassification of Goods for Fraud Detection

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
0
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
3
1

Relationship

0
4

Authors

Journals

citations
Cited by 4 publications
(2 citation statements)
references
References 6 publications
0
0
0
Order By: Relevance
“…From four datasets (ranging in size from 4,000 to 12,000) obtained through Electronic Data Interchange (EDI) files, the results showed a top-10 precision that varied from 43% to 70% on the best sentence embedding model Universal Sentence Encoder (USE), along with variations of the Bert model [2]. Closer to our work, Spichakova and Haav [7] introduced an innovative way that allows for the prediction of HS6 codes using a combination of two different measures on the Bill of Lading Summary 2017 dataset. First of all, using the Doc2Vec [8] embedding a cosine similarity measure of sentences describing a product is calculated, displaying the most similar text sentences associated with an input and the corresponding HS6 codes.…”
Section: Related Worksupporting
confidence: 57%
“…From four datasets (ranging in size from 4,000 to 12,000) obtained through Electronic Data Interchange (EDI) files, the results showed a top-10 precision that varied from 43% to 70% on the best sentence embedding model Universal Sentence Encoder (USE), along with variations of the Bert model [2]. Closer to our work, Spichakova and Haav [7] introduced an innovative way that allows for the prediction of HS6 codes using a combination of two different measures on the Bill of Lading Summary 2017 dataset. First of all, using the Doc2Vec [8] embedding a cosine similarity measure of sentences describing a product is calculated, displaying the most similar text sentences associated with an input and the corresponding HS6 codes.…”
Section: Related Worksupporting
confidence: 57%
“…Isto, alinhado com o grande volume de notas fiscais emitidas, dificulta o processo de auditoria e investigação de fraudes por parte dos órgãos competentes. Esse problema não é exclusivo do Brasil, em 2017, por exemplo, o Tribunal de Contas Europeu comunicou que as formas de evasão de pagamento amplamente aplicadas são a subavaliação, classificação errada ao mudar para uma classificação de um produto com alíquota mais baixa, e a descrição errada das mercadorias (Spichakova & Haav, 2020).…”
Section: Motivaçãounclassified