The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Ulysses – Smart analysis on e-Polls


Added by:   OECD analyst
Added on:   05 Aug 2026
Updated by:   OECD analyst
Updated on:   05 Aug 2026

“Ulysses – Smart analysis on e-Polls” is an AI algorithm that analyses citizens' comments on e-polls about legislative bills available. The algorithm identifies the main arguments for and against each legislative bill in progress, grouping them. As all legislative bills are available for e-polls, there is a huge number of citizens contributing opinions. In addition, quantitative indicators are not sufficient to effectively identify exactly the items of articles accepted or rejection by citizens.

Name in original language

Ulysses - Análises inteligentes em enquetes

Initiative overview

The Brazilian Chamber of Deputies defined that all bills presented by parliamentarians are systematically made available to receive citizens' contributions through e-polls. There are about 16,000 proposals per year, and some of them receive more than 1,000,000 suggestions/comments from citizens during the weeks of greatest discussion on the topic. 

It was observed that merely knowing the percentage of citizens who were for/against a legislative bill is not sufficient to improve the legislative text. Therefore, it was necessary to add an analysis of a free field in the e-poll through which citizens are invited to leave their comments and opinions. A human full analysis on these opinions is practically impossible, due to the large number and the timeliness with which parliamentarians need to position themselves in debates regarding those bills. 

The "Ulysses – Smart analysis on e-Polls" algorithm identifies the main arguments for and against each legislative bill in progress, grouping them and showing how distant the opinion groups are from each other. The analysis is semantic, which enables deep analysis of opinions, identifying situations of irony included in the texts. In addition to results presented in text format, the algorithm also presents a graphical view of citizens' opinion groups.

Results, outcomes and impacts
More accurate and timely analysis of the arguments for and against a legislative bill, which makes it possible to identify: 

  • Which part of the legislative text is most accepted or rejected by citizens; 
  • Whether there is any problem with society's interpretation of the legislative proposal, which may have resulted from disinformation.

Challenges and lessons learned

  • Qualitative analyses of large volumes of data need to be incorporated into analyses on content received from citizens;
  • This type of analysis is an important mechanism for identifying the impact that disinformation has on citizens' opinions

About the policy initiative


Category:

  • AI policy initiatives, programmes and projects

Initiative type:

  • AI use cases/projects in the public sector

Status:

  • Active

Start Year:

  • 2022

Target Sectors:


OECD AI Principles:


Other relevant urls: