The need to visualize network data becomes increasingly crucial, as it appears in numerous domains like web visualization, computer network monitoring, or social studies. Graphs and matrices are two well-known techniques to visualize network data. Depending on which task has to be carried out on the data, each of them offers complementary useful properties. This paper describes an attempt to combine some advantages of both techniques within a single representation. The proposed visualization, called Mat?Graph, is based on the adaptation of a graph to fit a matrix layer. This approach provides a new trade-off between advantages and drawbacks of both techniques. The matrix-based adapted graph has been used to visualize electronic message interactions.