The hackathon held by the WIN project was an online challenge of 6 weeks in autumn of 2024. The hackathon was promoted via the ESSnet WIN social media channels, NSIs involved in the project, and other external stakeholders, such as national statistical associations across Europe and international organisations. This led to the registration of 10 teams originating from different organisations registered for the challenge.
The challenge to be performed was described on the CROS portal pages of the project. In brief, the challenge was to develop open-source software, to be published under an open source license on a public GitHub repo, to score a given dataset of 4000 URLs of enterprises in 4 different countries (NL, AT, PL, DE). For each country 1000 URLs were available. The binary variables that had to be derived from the websites were e-commerce and social media use. The latter variable had subcategories Facebook, LinkedIn, X, Instagram, TikTok and YouTube.
The dataset was derived from public available entries on maps spread over different regions in the respective countries and varying in activity of the enterprises. The data was deduplicated and a sample was taken to arrive on the dataset for the challenge. A subset of 100 URLs per country was manually labelled by the project partners to arrive at a (secret) validation set to decide on the winner(s).
Two winners have been chosen. Their results can be found here:
Roshna Omer (UNHCR)
Enhanced Social Media and E-commerce Detector aka:
https://github.com/RoshnaOmer/win-hackathon/Riccardo Corradini, Rita Lima (ISTAT)
Freesoftwdreamer team:
https://github.com/freesoftwdreamer/Web-Intelligence
One of the winners, Riccardo Corradini, presented their work in the WIN session at the NTTS 2025.