Learning system dynamics from the observations of state transitions has many applications in bioinformatics. It can correspond to the identification of the mutual influence of genes and can help to understand their interactions. A model can be automatically learned from time series data by using methods like Learning from interpretation transition (LFIT). This method learns an exact model if all transitions of the systems are used as input. However, in real biological data, such complete data sets are usually not accessible and we have to learn a system with partial observations. Usually, biologists also provide with a priori knowledge about the system dynamics in the form of temporal properties. When building models, keeping critical properties valid is one of the major concerns and model checking plays a role in the verification of such desired properties. Our research aims at providing a model checking approach to revise logic programs thanks to temporal properties. In this paper, as a first step, we propose a method that can exploit reachability properties to fit such a model.