Senior software developer in the Perfomance in Data modeling team @ MathWorks

GitHub Scholar LinkedIn Twitter Google+
Home Teaching Publications Tools Projects


Maven-miner: Mining Maven Central and creating the Maven Central dependency graph of (Main developer)

Maven miner is a set of java tools aiming at, programmatically, resolving all Maven dependencies hosted in the Maven central repository, then, storing them into a graph database. We rely on Neo4j, a well-known graph database, to persist the maven dependency graph. The resulting dataset is publicly available in Zenodo. DOI

ATL-MR: Model Transformation with ATL on MapReduce (Main developer)

ATL-MR is a prototype tool for running complex ATL transformation in the cloud using Hadoop MapReduce. ATL-MR is implemented on top of an extended ATL VM that can be found here. Coupling ATL-MR with the extended VM has proved a good performance, especially in terms of execution time. Interested in testing ATL-MR? A ready-to use docker cluster is at your disposal.

NeoEMF: Scalable Model Persistence on NoSQL Stores (Contributor)

NeoEMF is a multi-backend for the manipulation and persistence of EMF-based models in NoSQL databases. Currently, NEOEMF is implemented in three different NoSQL stores, Graph, Map, and Column. Each one is well-suitable for a specific modeling scenario.

NeoEMF Instantiator (Contributor)

A pseudo random instantiator for EMF models stored in NeoEMF. This work is adaptation of the original random generator for XMI models.

EMFtoCSP: Scalable Model Persistence on NoSQL Stores (Contributor)

EMFtoCSP is a tool for the automatic verification of UML or EMF models annotated with OCL constraints. It can check automatically several correctness properties about the model, such as the satisfiability of the model or the lack of contradictory constraints.


ATL Zoo Benchmark

This benchmark selects a subset of the transformations in the ATL Transformation Zoo based on their quality level and usage in real-world applications. It specifically includes only transformations that may be used in production environments. The benchmark comes with a deterministic model instantiator and a default generation configuration.

OpenBIM Benchmark

This benchmark includes a model validation and transformation use cases in the context of construction industry. It comes also with a set of EMF-compliant BIM (Building Information Model) models with increasing size going up to 1Gb. For confidentiality purposes, only obfuscated input models have been made publicly available.

ITMFactory Benchmark

This benchmark contains two transformations and a set of queries, each addressing a different phase in a model-driven reverse engineering process. The use case for this benchmark is taken from the Eclipse MoDisco\footnote{\url{}} project. Thanks to the MoDisco Java discoverer, we are able to extract Java models up to 1.3GB, that conform to the Java metamodel defined in MoDisco. A systematic way to reproduce input models for this use case is also described.