What is it about?

When citizens want to know what happened at a local city council meeting, they are usually faced with lengthy, dense, and highly formal documents. To help solve this, we created "CitiLink-Summ", a new dataset built from 120 official meeting minutes across six Portuguese municipalities. A team of linguistics experts carefully broke these documents down and hand-wrote 2,880 short summaries, each focusing on a specific subject discussed during the meetings. We then used this dataset to test how well modern AI models can automatically generate these summaries.

Featured Image

Why is it important?

While official meeting minutes are essential for government transparency and accountability, their sheer length and complexity make them difficult for the average person to navigate. Automatic text summarization can fix this by giving citizens the main subjects without requiring them to read the full text. However, AI models need high-quality examples to learn from, and this kind of data for administrative texts in European Portuguese has been very scarce. By providing the first benchmark dataset of its kind in this language, we are laying the groundwork for tools that can ultimately make local decision-making more accessible to the general public.

Perspectives

Working on this project really highlighted the gap between public information and actual public accessibility. Municipalities produce huge amounts of text to ensure transparency, but if citizens cannot easily digest it, that transparency is somewhat lost. I found it fascinating to explore how we can bridge that gap using AI. One of the most rewarding, and challenging, parts of the process was seeing just how much complex reformulation our linguistics team had to do to summarize these documents accurately. It showed me that teaching AI to understand highly formal, administrative Portuguese isn't just a technical challenge, but a crucial step toward better civic engagement.

Miguel Marques
Universidade da Beira Interior

Read the Original

This page is a summary of: CitiLink-Summ: A Dataset of Discussion Subjects Summaries in European Portuguese Municipal Meeting Minutes, April 2026, ACM (Association for Computing Machinery),
DOI: 10.1145/3774904.3792945.
You can read the full text:

Read

Contributors

The following have contributed to this page