What is Serenata de Amor?
Serenata de Amor is an open project using artificial intelligence for social control of public administration. It was launched in September 2016 as part of the Data Science for Civic Innovation Programme from Open Knowledge Brasil. The project tracks government spendings and makes open data accessible for everyone. It started by analyzing data from the Chamber of Deputies (Brazilian lower house) and expanded to the Federal Senate (Brazilian upper house).Serenata de Amor is part of Open Knowledge Brasil’s civic innovation programme. The project is maintained by a distributed community of contributors and runs entirely on open source software.
The two main components
Rosie
An artificial intelligence robot that reads congressional expense receipts, applies machine learning classifiers, and produces a dataset of suspicious reimbursements. Rosie also tweets her findings at @RosieDaSerenata.
Jarbas
A Django-powered web dashboard and REST API that serves Rosie’s findings to the public. Users can browse congresspeople expenses and get details about each suspicion — the starting point to validate a finding.
Mission
The project’s mission is to empower citizens with data. Rather than focusing on smart cities and surveillance, Serenata de Amor focuses on smart citizens, accountability, and open knowledge.What Rosie analyzes
What Rosie analyzes
Rosie reads reimbursements from the Quota for Exercising Parliamentary Activity (CEAP) — the public fund that congresspeople use to reimburse work-related expenses. She flags reimbursements that appear suspicious based on several machine learning classifiers.
How findings reach the public
How findings reach the public
Once Rosie produces a
suspicions.xz dataset, Jarbas imports it and makes every suspicion searchable via a web dashboard and a REST API. Citizens can filter by congressperson, date, category, and suspicion type. Rosie also posts individual suspicious reimbursements on Twitter, tagging the congressperson involved.Geographic scope
Geographic scope
The project’s main implementations focus on Brazil. The Chamber of Deputies and Federal Senate are the primary targets, though the architecture is designed to be adaptable to any country that offers open data.
The Twitter bot
Rosie is active on Twitter as @RosieDaSerenata. When she identifies a suspicious reimbursement, she tweets about it — engaging citizens directly and inviting them to interact with their representatives.The serenata-toolbox package
Theserenata-toolbox is a pip-installable Python package that handles downloading and versioning the raw datasets from the Chamber of Deputies and Federal Senate. Rosie depends on it to fetch up-to-date reimbursement records before running her analysis.
Open source and community
GitHub
The main repository hosting Rosie and Jarbas. Contributions are welcome.
Community notebooks
Experimental Jupyter notebooks maintained by the community for research and exploration.
Discord
Join the Open Knowledge Brasil Discord to discuss the project, report bugs, and propose improvements.
Support
Support the project through the recurring crowd funding campaign on Apoia.se.
Operation Serenata de Amor has expanded into new projects under Open Knowledge Brasil. Rosie and Jarbas receive updates less frequently than in earlier years. If you are looking for a highly active community, consider also exploring the Querido Diário project.