Machine learning-powered risk scoring: ML algorithms
that had been trained on historical data related to tax
fraud were deployed to identify relevant risk factors.
Based on the criteria identified by the ML algorithms,
the organization’s analysts configured business rules
to flag and prioritize important risk factors that were
previously not considered. NEXYTE then automatically
scored and prioritized audit candidates based on risk.
Investigation workspace: NEXYTE presented
discrepancies between reported and actual income to
surface audit candidates. AI engines were then applied to
extract and analyze data from images and texts in seller
postings of these audit candidates and infer relationships
between sellers in the same business. These sellers
reported income separately to evade taxes. A visual link
analysis graph was automatically generated to enable
investigators to identify suspicious seller networks.
Unstructured data analysis: NEXYTE surfaced online
posts made on multiple online platforms and analyzed
structured and unstructured data, including videos.
NEXYTE’s video analytics engine was applied to extract
sales indications from recorded live streams to flag potential
unreported sales. Information from videos further provided
indications regarding potential discrepancies between
reported and actual sales.
NEXYTE allowed investigators to proactively surface sellers
and link between them and their tax declarations, as well
as to detect new online shops more easily. NEXYTE’s
investigation workspace enabled intuitive data exploration
using visual widgets, filters, and tags.
NEXYTE –Business intelligence dashboard
NEXYTE – Risk scoring
NEXYTE Investigation Dashboard
NEXYTE – Entity extraction from text
NEXYTE – Link analysis