i have read paper about graph convolutional networks
since it a document classification task, is it important to choose term weigthing method like tf-idf. the default term weighting of CORA dataset is using binary weighting, so maybe it will be better if i transform it to tf-idf maybe?
or maybe there is no difference on performance between binary and tf idf? can you explain why it is important or not?
If you could, can you use any paper references to explain that?