Prototype and Processbook added.
The experiment result on real data added.
We select 35 graphs in SuiteSparse Matrix Collection to test our conclusion on real data. Due to the limited number of real data, we trained our data on randomly generated synthetic data. The outcome of this experiment is not good since synthetic data do not have the complex artificial features like it in real data. However, the outcome of this experiment cannot lead to the conclusion that topological features cannot represent layout. Further experiment can be conducted when enough real data are available for training.