Browse Prior Art Database

Detecting possibly fraudulent operations by analyzing anomalies in invoice chain Disclosure Number: IPCOM000250427D
Publication Date: 2017-Jul-13
Document File: 1 page(s) / 26K

Publishing Venue

The Prior Art Database


Solution presented allows for easy and quick detection of suspicious flows of goods. This is intended to be used by tax and law authorities to detect possibly fraudulent operations.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 86% of the total text.

TITLE: Detecting possibly fraudulent operations by analyzing anomalies in invoice chain Summary:

Idea is to analyze how, for particular good type, chain of invoices look like. And based on that decide if it looks "normal" or "strange". Description: As mentioned in summary, we propose to analyze invoice chains for particular goods. This can be easily done using one of many machine learning methods, as the output we will point out chains that stand out from norm/average. Invoice chain is a sequence of invoices connected via certain good item and seller- buyer relationship, see figure.

Figure - invoice chain

Embodiment: Imagine a factory that manufactures cigarettes. Then those are sold to some wholesaler who is selling them to small shops. This is the most typical case, so in this particular situation invoice chain has three elements: manufacturer-wholesaler-shop (-customer, who is not always invoiced). Of course it may happen that for instance there are two wholesalers in the chain (regional and local for example). Even with very simple analysis of chain length one can tell, that if that length is, say 12 it looks suspicious. Another type of analysis that may be performed is that chain start and/or end nodes look strange - there are not that many (not many in terms of computers) cigarette (or any other particular good) manufacturers (or importers) and if cigarettes out of the sudden start to seemingly originate from previously unknown source it may also be treated as anomaly...