COFE Datathon

1. WHAT IS THE COFE DATATHON?
The COFE Datathon is a data mining competition organised by the European Commission in the context of the Conference on the Future of Europe. The aim of the datathon is to foster the analysis of the Conference open data set to discover new and rich insights around the public contributions submitted by citizens to the digital platform.

2. WHEN AND WHERE WILL IT TAKE PLACE?
The COFE Datathon will start on the 1st of March 2022 and officialy end on 30 March 2022. During this period, teams participating in the datathon will be able to use their own tools to perform the analysis and submit their results.

On 30 March 2022, there will be a closing event, with a pitch session where each team will have 10 minutes to explain their approach and main findings, and a closing ceremony.

3. WHO CAN PARTICIPATE?
The COFE Datathon is a competition open to participation to any team from anywhere in Europe, composed of a maximum of five members. Teams can be composed by individuals of private companies, universities, research institutes, public entities, associations, etc. The competition is not open to staff working in the EU institutions and agencies.

4. REGISTRATIONS
Registrations to the datathon will be open until 27 February 2022. Participating teams will need to fill in the registration form indicating: the name of the team, member’s names, nationality, and contact email. The registration form is available at: https://ec.europa.eu/eusurvey/runner/cofe-datathon

Teams who plan to attend the online kick-off meeting (1st of March 2022) should contact the datathon organisers (EU-CITIZEN-ENGAGEMENT@ec.europa.eu) before the 27 February 2022 (midnight CET).

5. WHAT ARE THE ANALYTICAL CHALLENGES?

The datathon is open to any contribution that provides analysis of the collected data and interesting insights. We propose 5 challenges as examples of relevant tasks:

• Challenge 1: Keyword extraction from proposals (named “ideas” on the COFE digital platform) - represent an idea or a set of ideas by a set of most representative keywords.

• Challenge 2: Idea clustering - group ideas expressing similar notions.

• Challenge 3: Main proposition detection - extract from ideas the most often mentioned specific propositions.

• Challenge 4: Stance detection - detect if a comment about an idea is in favour or against of the corresponding idea.

• Challenge 5: Open task - this task can be summarized as "If you want to study problems not listed above, you are free to do so by participating to this task".

Each challenge is independent from the others. Detailed technical information will be provided on the start of the datathon.

6. DATA ACCESS
The dataset used in the datathon will be the one publicly accessible from the COFE digital platform. All data will be anonymised – usernames and user geolocations must be replaced by anonymous alphanumerical ids.

The datasets for analysis will be provided to all teams in the beginning of the competition (1st March 2022) by the COFE Datathon organisation via email / ftp.

7. EVALUATION
A set of teams will be invited to present their findings related to each of the five challenges at the datathon closing event. The selection will be based on a qualitative assessment based on criterial such as multilingualism, methodology, scalability, reliability and clarity of the presentation of results. Detailed guidelines will be provided on the start of the datathon.

Each challenge is independent and will be evaluated separately (i.e., teams can choose how many and which challenges they want to address).

The evaluation will be performed by a committee of data specialists and analysts chaired by the European Commission’s Joint Research Centre.

8. WILL THE OUTCOMES OF THE DATATHON BE LINKED TO THE CONFERENCE?
Contributions from participating teams will be made available on the Conference’s digital platform in time of the Conference’s final Plenary session.

Selected teams based in quality criteria will be invited to present their findings in the datathon closing event.

Based on the interest and outcomes of the Datathon we envisage organising at a later stage this year a more classical shared task that will be co-located with a main scientific conference related to data science.

9. IMPORTANT DATES

1st February 2022: Announcement of the COFE Datathon competition. Registration period is open.
27 February 2022 (midnight CET): Close of registrations.
1st of March 2022: Launch of the COFE Datathon with the kick-off meeting. The datathon starts with an online event presenting the concept and explaining the different challenges. Datasets are provided to the teams.
20 March 2022: Deadline for the teams to submit contributions.
30 March 2022: Closing event with solution presentation session.

10. DATATHON OUTCOMES

10 contributions were received from 7 teams, addressing 3 of the 5 challenges proposed: