Fight Cybercrime With Bloom: The 5-Minute Interview With Bertrand Provost


“I was amazed by the quality of Bloom and the ability to travel the graph. It makes for quite a clear and user-friendly display,” says Bertrand Provost, Senior Data Scientist at La Poste.


Meet Bertrand Provost, a Senior Data Scientist at Le Groupe La Poste, a company that operates postal services, banks, insurance, and more. And as France’s biggest local commercial network with 1.3 million daily customers – and delivering 18 billion items annually (!) – it’s not difficult to see how the organization may face some data challenges. Enter graph technology.

We got to catch up with Bertrand at the Paris stop of our inaugural GraphSummit tour, and learned more about how he uses Neo4j to meet his objectives.

Check out the latest 5-Minute Interview installment below.

Please introduce yourself.


Bertrand Provost: My name is Bertrand Provost. I’m a Senior Data Scientist at La Poste, which is quite a big company owned [partially] by the French government. My role is to mitigate fraud and to fight cyber criminality. I use Neo4j to model the journey of our user on the bank website, to be able to detect fraud or suspicious journeys.

What are some surprising results you’ve seen from using Neo4j?


Bertrand Provost: I was amazed by the quality of Bloom and the ability to travel the graph. It makes for quite a clear and user-friendly display, with Bloom.

Do you have any advice for someone getting started with Neo4j?


Bertrand Provost: I’d say that you need the help of professional services to set up a model to – above all – ingest the data, to get the data into the database. Otherwise, it’s very frustrating and sometimes daunting, if you know what I mean.



What do you think is in store for the future of graph technology?


Bertrand Provost: As a data scientist, I’m… designing an algorithm to model NLP – natural language processing. Now that I’m aware of all the capabilities of Neo4j, I think I’m going to use Neo4j to model and design my new NLP algorithm, because each “word” can be linked to another one. There is a relationship between two words, and in the word phrase, in the word sentence.

I think it’s going to be a good help to me to prepare my dataset using Neo4j and export the relationship created by data by Neo4j to feed my new network. To be able to make predictions on price and sentencing and such.



Learn how to harness the power of graph databases for your enterprise fraud detection efforts with this white paper, Fraud Detection: Discovering Connections with Graph Databases.

Download the White Paper