Applied Research Unit (BIARU)

A classical example of machine learning, in which properties of second-hand products are exploited to estimate a future selling price.

The pricing engine serves to support resellers in their decision making. Challenges in the project include dealing with small datasets as well as incorporating domain knowledge and information hidden in unstructured data

The Flemish Audit Authority (Vlaamse Auditautoriteit) is tasked with auditing projects funded by European structural funds with Flemish co-financing. Current audits are delayed two years after expenses, limiting opportunities to recover rejected costs. Additionally, audits must be completed within a constrained timeframe from April to November, leading to frequent outsourcing at high costs with limited benefits. Outsourced auditors are often inexperienced, requiring extensive oversight by the Authority, which undermines the outsourcing's intended efficiency.

This project, led by Deloitte Belgium, seeks to revamp the audit process both methodologically and technologically to improve effectiveness, reduce costs, accelerate timelines, and protect European funds from losses. To reach this goal, BIARU contributes to the Process Monitoring Cockpit component of the process, addressing research challenges dealing with long-term and cross-instance dependencies in process executions, and the visualization thereof.

More information can be found here.

Legislative bodies produce a large number of texts that describe updates in laws, regulations, and procedures. As citizens and companies, we are expected to know every law and regulation to abide by them.

Via Natural Language Processing capabilities, we identify and highlight relevant legislation, using examples of previous legislation deemed relevant. As a result, the number of texts that need to be browsed manually is drastically reduced.

Research challenges include dealing with unbalanced datasets and adequate processing of unstructured information.

With the building and construction sector under pressure, previous experience seems less and less useful to predict market fluctuations.

We examine the use of Machine Learning to extract valubale information from market surveys to predict market fluctuations.

Research challenges include dealing with small datasets that contain disruptive periods caused by the COVID-19 pandemic.

Research on the jurisdiction, especially when analyzing the motivation of judges and lawyers, is characterized by a large amount of manual work that requires the researcher to read and code as many juridical documents as possible.

In this project, we automate the information extraction pipeline by extracting relevant information from verdicts. The research is complicated by the use of juridical jargon, incomplete and unstructured motivations, and references to precedents that are not necessarily well-defined.

A collaboration with the Flemish Institute for Logistics, funded as a COOCK project by VLAIO.

The BILOG projects translate the academic and practical knowledge of Business Intelligence to the context of companies in the logistics sector. The project produced a step-by-step manual for companies that want to get started with Business Intelligence. Rather than reinventing the wheel, we create specific guidelines, templates, and examples that guide companies from discovering potential projects over designing data structures and tool selection to creating effective visualizations.

Together with PXL and funded by VLAIO, we help companies to discover potential use cases for Artificial Intelligence within their own company.

Via hackathons, exemplary use cases and a hackathon, we make AI tangible to companies that do not necessarily possess the required knowledge.on Artificial Intelligence and help them understand these new technologies' benefits and pitfalls. We show how a company can start with AI with limited investments and grow from there.

Marketers try to drive customer engagement via targeted marketing actions.

Exploiting the capabilities of process mining to discover the customer journey and how past marketing actions influence it, we develop an algorithm that supports marketers in selecting the right actions.

The algorithm assigns a customer or a lead to a customer profile and suggests marketing actions based on what has proven to be successful in the future.

Thanks to the support and financial resources from the Flemish Region and Europe, we are developing a new Training Course, “Successfully transform Strategic Transformations into AI projects,” in collaboration with INNO.COM and SEIN.

The ability of organizations to translate a business challenge into a technological solution dictates the successful adoption of AI in Flanders. One element is key: always take the business’ strategic needs and opportunities as a base to draw up a fitting and concrete AI project while including a clear-cut business case.

This training course will focus on two target audiences who are both crucial in successfully adopting AI. For each of these audiences, we’re developing a designated module. These modules will complement each other and transfer essential knowledge and skills to make an AI transformation happen.

As the amount of data that has to be managed by companies is growing exponentially, data warehouses have become an invaluable asset. In this project, The Agile Information Factory and BIARU join forces to bring data warehouse automation to the next level by removing manual steps from the configuration process.

This TETRA project aligns the operations of Flemish freight carriers with IT-driven market evolutions through the innovation of their business processes and supporting information systems. We developed a digitalization scale based on business process management, which allows freight carriers to assess their readiness for e-CMR and Information-on-Demand innovation projects and execute the necessary preparatory actions.