Labour inspectors face a daunting task every day. Beyond identifying and deterring violations of labour laws, they must offer guidance, provide technical advice, and promote good practices to foster healthier and more compliant workplaces. This work involves sifting through thousands of cases to spot violations – a time-consuming process that can often miss critical cases and wastes valuable resources. 

That’s where the Matrix of Intelligence and Risk Assessment (MIRA) comes in. Developed for the Albanian Labour Inspectorate and Social Services with the support of the International Labour Organization (ILO) and funding from the European Union, MIRA revolutionizes how inspections are conducted. Moving away from traditional tools like the “Matrix of Penalties” and manual red-flag assessments, MIRA integrates advanced data and case management with risk assessment capabilities powered by data mining and machine learning.

Eljo Muçaj, Albania’s Chief Inspector, emphasized the need for transformation: “Traditional approaches were overdue for an overhaul. The solution lay in modernizing our tools and embracing technology to make inspections more effective.”

Through its machine learning risk assessment capacity, MIRA offers a breakthrough solution that revolutionizes labour inspection planning. The system uses simultaneously nine different advanced algorithms with historical inspection data to detect cases of undeclared work or other labour law and OSH violations more accurately while substantially reducing analysis and reporting time.

Illustrative image (internet)

MIRA is a comprehensive tool that brings together two key components: a case and data management system and an advanced risk assessment tool that runs on data mining and machine learning. 

One of MIRA’s key benefits is its real-time access to the latest laws and regulations. Inspectors have the most current information at their fingertips, avoiding the risk of outdated or inconsistent practices. Together with standardized workflows, you’ve got a system that ensures every business is treated equally. This consistency not only saves time but also builds trust so that businesses can feel confident knowing inspections are fair and transparent.

Harnessing Data Mining and Machine Learning

MIRA’s structured data collection system and predictive analytics are transforming the way labour inspections are planned and conducted. By collecting and organizing detailed data, MIRA builds a comprehensive repository that labour inspectors can use to identify patterns and risks. But it’s the advanced technology behind MIRA – data mining and machine learning – that truly sets it apart.

Data mining

Data mining is MIRA’s first step in making sense of the vast amounts of information it collects. Combing large datasets, MIRA helps to uncover hidden trends, correlations, and patterns that might otherwise go unnoticed. 

MIRA helps inspectors understand where to focus their attention. It acts like a spotlight, illuminating areas of concern within labour relations and occupational safety and health. Without data mining, this wealth of information would remain buried in spreadsheets and reports, leaving inspectors reliant on intuition or incomplete data. 

Machine learning 

MIRA works exceptionally well because it conducts complex analysis of relevant factors captured in historical data using nine machine learning algorithms. 

For instance, machine learning can predict potential businesses and spot where undeclared or underdeclared work might happen based on the business’s size and type, location, industry, seasonal variation, work hours, time of work and compliance history. The implementation of the machine learning showed a 30 per cent improvement in predicting undeclared and underdeclared work compared to previous red-flags assessment methods. 

Albana Kuka, the Head of the Risk Analysis Sector, stated that “…approximately 68 per cent of identified informal employment cases are part of planned inspections, a noticeable improvement compared to unplanned inspections. Automated planning has reduced errors and expedited the process, allowing inspectors to focus on identifying and addressing the most critical violations.”

MIRA’s predictive power is impressive – 70 per cent of inspections are planned based on its risk assessment and data-driven insights. However, 30 per cent of inspections are random inspections, selected for reasons other than those identified by MIRA. This balance serves two key purposes: collecting fresh data to improve accuracy of predictions and adapt to changes in the labour market; and ensuring accountability by covering all businesses by random inspections. 

Overcoming Challenges and Bias

Adopting AI-driven tools like MIRA wasn’t without challenges. Inspectors had to adapt to the idea of relying on machine predictions, which sometimes challenged their assumptions about where violations occur. Traditional inspection planning favours certain industries or regions based on anecdotal experiences or assumptions. Irida Qosja, Director of Labour Inspections, noted that “Initially, the system was perceived as an additional burden on their daily workflow. However, step by step, its benefits became evident. The transparency in the inspection process, turned out to be a crucial element in enhancing the credibility and effectiveness of inspections. MIRA shifted the focus from the inspectors to the process itself, placing it at the core of operations.” 

MIRA scrutinizes variables that labour inspectors would not give a second thought to, thus spotting behaviours or changes in behavioural patterns faster and more accurately than labour inspectors or their analysis team. Relying on data rather than personal judgment helps reduce the risk of subjective bias, ensuring decisions are objective and consistent.

Unlike static systems, MIRA learns and improves over time. With over 12,000 inspections conducted annually in Albania and data collected for more than 100 variables per inspection, MIRA’s machine learning and predictive capabilities grow stronger. It adapts to new trends and scales effortlessly, handling larger datasets without requiring additional resources. This means MIRA doesn’t just help inspectors work smarter today – it keeps evolving to meet future challenges. 

The Future of Labour Inspections

As the Albanian Labour Inspectorate continues to refine its processes, MIRA stands as a model for how technology can transform labour inspections. It empowers inspectors to focus on their broader mission: creating workplaces that are safer, fairer and better for everyone.