Ricardo Brancas

PhD candidate in Computer Science · Lisbon

Email Scholar ORCID GitHub LinkedIn
Ricardo Brancas

About

I'm a Computer Science PhD student in the CMU Portugal Affiliated PhD Program, advised by Vasco Manquinho at Instituto Superior Técnico and Ruben Martins at Carnegie Mellon University. My research helps programmers — and non-programmers — get through their daily work, by leaning on program synthesis and program repair. I'm also drawn to formal methods and compilers. Always happy to talk about any of it.

Selected Publications

2026

ProDebug: An Automated Debugging System for Prolog — TPLP / ICLP 2026

Ricardo Brancas, Vasco Manquinho, Ruben Martins

To appear in Theory and Practice of Logic Programming; presented at the 42nd International Conference on Logic Programming.

2026

What Bugs Do Prolog Students Write? An Empirical Taxonomy and Data-Driven Mutation Framework — ICLP 2026

Ricardo Brancas, Pedro Orvalho, Carolina Carreira, Vasco Manquinho, Ruben Martins

To appear in the Technical Communications of the 42nd International Conference on Logic Programming (EPTCS).

2026

Can Automated Feedback Turn Students into Happy Prologians? — ICLP 2026

Ricardo Brancas, Pedro Orvalho, Carolina Carreira, Vasco Manquinho, Ruben Martins

To appear in the Technical Communications of the 42nd International Conference on Logic Programming (EPTCS).

2025

Combining Logic and Large Language Models for Assisted Debugging and Repair of ASP Programs — ICST 2025

Ricardo Brancas, Vasco Manquinho, Ruben Martins

Introduces FormHe, a tool that pairs logic-based techniques with LLMs to detect and correct issues in Answer Set Programming submissions.

PDF arXiv Code
2025

CUBES: A Parallel Synthesizer for SQL Using Examples — Formal Aspects of Computing

Ricardo Brancas, Miguel Terra-Neves, Miguel Ventura, Vasco Manquinho, Ruben Martins

Extended journal version of the FASE 2024 paper, with an expanded evaluation on ~4000 SQL queries; the parallel synthesizer scales to 16 processes with super-linear speedups, and disambiguation pushes accuracy to ~60%.

DOI Code
2024

Towards Reliable SQL Synthesis: Fuzzing-Based Evaluation and Disambiguation — FASE 2024

Ricardo Brancas, Miguel Terra-Neves, Miguel Ventura, Vasco Manquinho, Ruben Martins

Introduces Cubes, a parallel SQL synthesizer using input/output examples and a user-guided disambiguation process — up to 60% accuracy with super-linear speedups on multicore.

PDF arXiv (extended) Code
2023

On Repairing Natural Language to SQL Queries — arXiv

Aidan Z.H. Yang, Ricardo Brancas, Pedro Esteves, Sofia Aparicio, Joao Pedro Nadkarni, Miguel Terra-Neves, Vasco Manquinho, Ruben Martins

PDF arXiv

Projects

ProFix

An ongoing project to support students taking Logic Programming courses. The first phase surveys students to map the most common difficulties when learning Prolog.

FormHe

A tool that helps users find and fix problems in ASP programs. A fault-localization module identifies which statements are off based on an exemplifying query; a repair module then suggests fixes via program synthesis.

Cubes

A parallel program synthesizer for SQL queries from input/output examples. Cubes extends sequential query synthesizers with new pruning techniques and a divide-and-conquer split of the search space.