OASC
team LESL · MIMathon Porto 2026
MIMathon Porto 2026 · team page

Team LESL.

Lea, Eliott, Sattisvar and Louis. Three organisations: Kereval, Dolfin, Askem. One shared method for harmonising messy open data against multiple open standards, validated on three Porto Open Data use cases.

EventMIMathon Porto 2026
Members4 across 3 orgs
Use cases shipped3 of 5 tracks
Methodpivot harmonizer pattern

Four people, complementary roles.

L
Kereval · interoperability validation

Lea

Software quality and interoperability validation at Kereval. Brings the rigour that turns "it runs on my laptop" into "this output validates against the official standard schemas".

E
Kereval · interoperability validation

Eliott

Software quality and interoperability validation at Kereval. Watches conformance gaps: missing required fields, wrong cardinalities, units mismatches, the things that break downstream platforms silently.

S
Dolfin · ontology language

Sattisvar

Author of Dolfin, the human-friendly ontology language used as the canonical pivot for every UC. Designs the model so that domain experts can read it and tools can compile it.

L
Askem · method and delivery

Louis

Method design and delivery at Askem. Architects the adapter-pivot-writer split, ships the pages and the harmonizer packages, keeps the work end-to-end reproducible.

Three companies behind the team.

Pivot language

Dolfin

A human-friendly ontology language. Backend-independent: the same Dolfin file targets graph databases, relational stores, document stores. Designed so domain experts can read it without training.

dolfin.fr ↗

Interoperability validation

Kereval

Software quality and interoperability testing. The team that turns standards conformance from a claim into a verifiable property.

kereval.com ↗

Method and delivery

Askem

End-to-end delivery of data harmonisation projects: method design, pipeline implementation, documentation, training. Built on Dolfin and on open standards.

askem.eu ↗

What we built at MIMathon Porto 2026.

Three Porto Open Data tracks, three different standards positions (gap, align, partial), all following the same pivot-and-writers method. Each page links the canonical Dolfin model, the source data, the harmonized outputs, and the full source code.

Common pattern

Each use case lifts source data to a Dolfin canonical model, then derives every output format from it. New source: one adapter. New consumer: one writer. The full method is documented as a single reusable skill, applicable to any harmonisation project beyond the MIMathon.