Nadia Boukhelifa, a data visionary

Nadia Boukhelifa received her secondary school degree – with honours – in Algeria, where she ranked among the top 20 students. After receiving a national scholarship, she decided to pursue her studies abroad. She chose England, a country known for its multicultural identity and excellent scientific schools where she could practise her English. She stayed 12 years.

From computer science to data visualisation

Nadia got a degree in computer science before deciding to get her doctorate at the University of Kent. She began working in the field of data visualisation, where her focus was on transforming often abstract data into the most effective visual forms to make the data more easily understood and interactive.

“It’s a little like turning a data table into a bar chart or a line graph,” explains Nadia, although, she adds, “data are generally complex and the tasks are often exploratory…”The aim is to better understand the data, explore them, show how they are connected and facilitate the related tasks.

Better understand the data and show how they are connected

From 2006, Nadia abandoned her plans to work in industry to instead become a researcher. She initially began in England, at the University of Leeds, where she worked on data integration, then moved to France, where she first joined Inria and later Telecom ParisTech for postdoctoral fellowships. Nadia was always interested in computers and had long been passionate about applications. She focused her work on the point where people and machines intersected with artificial intelligence. As her professional skills and network crossed borders, and she worked on a range of projects, from the history of Europe to resurfacing roads and visualising data uncertainty, which involved different stakeholders and user groups.

It was with this same mindset and motivation that she turned to INRA. The institute offered her a chance to work in a new field with new expert users, and the prospect was even more appealing as she had been working with several INRA teams for a while. Nadia was recruited in 2016 as a research scientist. She joined the Joint Research Unit for Food Process Engineering and Microbiology (AgroParisTech, INRA) at INRA’s Ile-de-France – Versailles-Grignon centre. This unit studies how to control the physical and biological processes that govern processing, from bioprocesses to people, with a view to producing knowledge and tools to aid the creation of products and bioproducts of high quality (sensory, nutritional, health and environmental).

Bringing people into the loop

Nadia is part of the MALICES team (Modelling, AnaLysis and knowledge Integration, for food and biological ComplEx Systems), which works to build models to represent these systems. Within this community which up until now has focussed on machine learning-based methods from AI and mathematical modelling, Nadia is working to “bring people into the modelling loop” through visualisation and interactive methods. More than just exploit experimental data, she seeks to include experts – from mathematicians crazy about coding to biologists who loved experimenting – in the modelling process to draw from their knowledge to better understand the data and guide the modelling. Their insights help her improve the models and ensure their relevance and purpose.

Facilitating discussion and collaboration

She has recently explored two complex models, one on fermentation and one on wheat fertilisation (like the AzoDyn model) using a visualisation tool she designed with researchers at Inria and continued developing at INRA, EvoGraphDice. The initiative, which experts were happy to try out on large tactile screens, revealed the advantages to this methodology: users were unanimously happy with the results, discussion and collaboration were facilitated, research was more accessible, and it provided a new way to manage conflicts between modelling criteria, such as how to optimise yield or reduce inputs.

These two-plus years at INRA have reinforced Nadia’s motivation, who says there is still “so much work to do” in terms of building, exploring and adjusting models using interactive visualisation while drawing from experts’ knowledge.

A vast programme – guided by the underlying issue of knowing whether interactive modelling is likely to generate new knowledge – for this new researcher, an extremely knowledgeable yet modest expert in data visualisation.