evidoria

← Back to browse

Good practice Imported

MiningWatch — Belgium's Data-Mining Tool for Targeting Social-Fraud Inspections

Belgium · Brussels · See the Belgium profile · See the Brussels profile

Evidence: Observational / pre–post Top 6% 87/100 · Ask Evidence Copilot about this practice

Since 2014, Belgium's federal social security service has used the MiningWatch data-mining tool to rank construction, cleaning and hospitality employers by fraud risk, lifting inspectors' detection success rate from 16% to 45% on a modest €200,000 annual budget.

16 %
Inspector detection success rate before MiningWatch (before 2014)
45 %
Inspector detection success rate using MiningWatch (after adoption)
200,000 EUR
Annual programme budget (annual)

Details

Maturity
Established
Promoter
Federal Public Service Social Security (SPF Sécurité Sociale), with the Sociale Inlichtingen- en Opsporingsdienst (SIOD) and four federal social inspection services
Period
2014–2026
Region (NUTS)
BE10
Keywords
social security compliance, undeclared work, labour inspection

Context

MiningWatch is a data-mining and predictive risk-scoring tool built in 2014 by Belgium's Federal Public Service Social Security to help the country's four federal social inspection services target undeclared-work investigations. It draws on the OASIS data warehouse, in place since 2001, which pools employer, worker, tax, posted-worker and land-register records, to rank companies in construction, cleaning and hotels/catering into red, orange, green and blue fraud-risk categories.

Objectives

The tool aims to help inspectors prioritise which employers to investigate for undeclared work, replacing less-targeted case selection with a risk-ranked list built from pooled administrative data.

Activities

A joint team of data miners and field inspectors reviews each proposed risk profile before it is loaded into MiningWatch and distributed as a ranked target list, so human judgement, not the algorithm alone, decides which profiles are acted on.

Results

According to the European Labour Authority's assessment, inspectors' average detection success rate rose from 16% before MiningWatch to 45% when using it to select targets, on an annual budget of roughly €200,000 for software licensing and service-provider costs.

Conclusions

The European Labour Authority has cited MiningWatch as a good-practice model for other EU labour inspectorates building similar risk-assessment systems.

Implementation
Indicative cost
Low (< €50k) — Runs on an annual budget of roughly €200,000 for software licensing and service-provider costs.
Time to results
Long (> 3 years) — Built in 2014 on top of the OASIS data warehouse established in 2001, and has run continuously since.
Staffing & skills
Dedicated core team of about ten data miners, IT staff and specially trained inspectors, Roughly 50 additional staff involved in applying the tool across the four federal inspection services, Joint review team of data miners and field inspectors who validate each risk profile before use

Conditions for success

  • Access to a mature, pooled data warehouse (OASIS) integrating employer, worker, tax, posted-worker and land-register records
  • Human review of every proposed risk profile before it is loaded and distributed as a target list
  • A dedicated recurring budget for software licensing and service-provider costs

Where it fits

Governance type
national/federal social security administration
Scale
national
Income level
high-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Replication kit

Reusable artefacts from this practice — as published by their sources.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful