The model presented in this talk is a computational model. It aims at representing knowledge by computational objects and at reasoning with the represented knowledge, i.e., at processing them by algorithms (philosophical or psychological aspects of knowledge will not be discussed). We first present the main properties a knowledge representation formalism should have and briefly survey graph or graphical models (in Computer Science). Then, we detail the model itself , which is graph-based in the following sense: knowledge is represented by labeled graphs and reasoning mechanisms are based on graph operations. The third part is devoted to the relationships of this model with logics and other computational models, especially the relational data base model and RDF/S. Finally, some tools and applications are mentioned.