Classification of budget amendments

Classification systems

Brazil - Senate

Use case ID: 070

Author: Federal Senate of Brazil

Date: 2 October 2024 

 

Objective: 

Intelligent and efficient classification of budget amendments into their programmatic functional category (Budget Unit, Function, Subfunction, Programme, Action), based only on their justification text.

Actors: 

  • Budget consultants
  • Parliamentarians who propose amendments to the budget
  • Lexor system: computerized system used by the National Congress to prepare amendments to budget laws

Prerequisites: 

  • Integration with the information system (Lexor)
  • Generated embeddings of the national register of budgetary actions
  • Trained artificial intelligence (AI) model for text embeddings

Scenario: 

  1. A user accesses the Lexor system.
  2. The user writes the justification text for the amendment to the budget.
  3. The user requests the AI-powered classification of the amendment.
  4. The system performs the embedding of the input text.
  5. The system performs a semantic search in the embeddings of the national register of budgetary actions.
  6. The system returns the most relevant programmatic functional classifications to the user, ordered by semantic similarity.
  7. The user selects the correct option from those returned by the system.

Alternate flows: 

  • If none of the options returned by the AI system is correct, the user can still choose a programmatic classification manually through Lexor.

Expected results: 

  • Efficiency in preparing amendments to the budget
  • Higher-quality classification of amendments to the budget
  • Reduction in review time for amendments by budget consultants

Potential challenges: 

  • Ensuring the AI system provides accurate and relevant responses 
  • Amendment texts written without a defined pattern
  • Similarities between budgetary actions

Data requirements: 

  • Embeddings of the national register of budgetary actions
  • National register of budgetary actions worksheet 

Integrations with other systems: 

  • exor system 
  • Analytics and reporting tools

Success metrics: 

  • Average response time
  • Percentage of successful hits in classification suggestions
  • Reduction in the number of incorrectly classified budget amendments

The Use cases for AI in parliaments collection is published by the IPU’s Centre for Innovation in Parliament as part of the Parliamentary Data Science Hub’s project to create guidelines for AI governance in parliaments.

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International licence. It may be freely shared and reused with acknowledgement of the author and the IPU. 

A use case describes how a system should work. It is used to plan, develop and measure implementation. A use case is not the same as a case study, which is a descriptive text of an actual project’s implementation. Please note that this use case is provided “as is” and neither the IPU nor the author accepts any responsibility for its use.

For more information about the IPU’s work on artificial intelligence, please visit www.ipu.org/AI or contact [email protected].