LLMs for legal-labor consultation: an undergraduate thesis presents a multimodal chatbot
Álvaro Henrique de Sousa Gouvea and Luís Guilherme Gaboardi Lins defended at FCTE/UnB an undergraduate thesis using Large Language Models to build a multimodal chatbot for legal-labor consultation.
Brasília, September 11, 2024
Workers who want to understand their rights on vacation, termination, or benefits rarely find a clear answer on the first query — and almost never in a single window. This is the friction the undergraduate thesis defended in September 2024 by Álvaro Henrique de Sousa Gouvea and Luís Guilherme Gaboardi Lins responds to with a multimodal chatbot, under the advising of Prof. Sergio Antônio Andrade de Freitas .
The work uses Large Language Models as the conversational core and adds multimodal layers — text, image, PDF — to handle the diversity of artifacts the questioner brings to the table: employment contract, receipt, document photo, WhatsApp message. In an AI and NLP setting, the proposal tackles a notoriously tricky slice: legal accuracy must coexist with colloquial language, and the model must know when to declare its limits.
The thesis grounds that conversation in a concrete artifact: a prototype that can be tried, critiqued, and improved. It is a typical contribution to a research line CEDIS has been consolidating since 2023, connecting LLMs to domains of significant public impact.
The full text — written in Portuguese — is available at the UnB Undergraduate Theses Digital Library.
About CEDIS: The Center for Studies, Development, and Innovation in Software (CEDIS), linked to the University of Brasília, researches and develops innovative software solutions.