portuguese-nlp

List of resources and tools developed with focus on Portuguese.

Article reproducibility LexIris-pt and LexBert-pt: Specialized Sentence Embeddings for Legal Similarity in Brazilian Portuguese.

This repository contains the official Python code and resources for the research paper: "Portuguese Automated Fact-checking: Information Retrieval with Claim extraction".

Honest benchmark: does DSPy beat a hand-written prompt, and at what cost? Manual vs. BootstrapFewShot vs. MIPROv2 on 3 real PT-BR tasks, reporting accuracy gains and USD cost.

End-to-End Python implementation of Muço’s (2025) corruption measurement framework. Combines NLP pipeline (regex extraction, Porter stemming, TF-IDF), PCA-based dimensionality reduction, and fixed-effects OLS to quantify institutional quality from Brazilian audit reports. Includes supervised learning robustness checks and LOO sensitivity analysis.

Portuguese split from MQA

Comparative study of 23 LLMs for Brazilian Portuguese sentiment analysis via in-context learning. Evaluates multilingual vs Portuguese-specialized models across 12 datasets. Code and data included.

Probing BERTimbau (Portuguese BERT) on B2W reviews: recovering vowel density via regression and classification.

Code and frozen results for diachronic lexical and semantic change in Brazilian Portuguese political discourse

Reprodutibilidade: deteccao de fake news em portugues brasileiro (generalizacao, atalhos de aprendizado e robustez). Qualificacao de mestrado e manuscrito submetido ao ENIAC 2026.

Code and data manifest for semantic drift in Portuguese financial disclosures

Code and data manifest for semantic drift in Portuguese financial disclosures

Code and frozen results for diachronic lexical and semantic change in Brazilian Portuguese political discourse

Article reproducibility Classification of the Conciliation Profile in Initial Petitions in the Brazilian Judiciary

Comparison of NLP approaches (BERTimbau fine-tuning, AutoML, and a Qwen 2.5 72B LLM) for detecting depressive symptoms in Brazilian Portuguese social media posts.

CurupiraIA: A Brazilian Portuguese hate speech detection model using BERT fine-tuning, inspired by folklore guardianship principles to protect digital communities from toxic content.