​Unlike neural networks, symbolic artificial intelligence algorithms simulate human reasoning by exploring a set of solutions to a given problem. The ExpressIF software platform developed by CEA Tech institute List, which uses deductive reasoning, was recently updated with a new type of reasoning known as "constraint satisfaction" reasoning, so that it can tackle even more complex problems.

Constraint satisfaction solvers are algorithms that effectively solve combinatorial problems. With fuzzy constraint satisfaction solvers, for example, priorities can be set and requirements and preferences introduced so that only certain constraints must be satisfied only partially. This is the kind of algorithm that was integrated into ExpressIF to give the platform greater flexibility in terms of how a problem must be stated (in natural language at that!) and to allow it to find more complex solutions.

This new reasoning component will position ExpressIF to address problems like scheduling, task allocation, positioning, and the annotation of items in images. It will be used on concrete tasks for EU research projects Micado (to analyze spectra and recognize isotopes using mass spectrometry) and Deephealth (to annotate medical images).