gcn_supervised transductive semi-supervised

"SEMI-SUPERVISED CLASSIFICATION WITH GRAPH CONVOLUTIONAL NETWORKS" 这偏论文是半监督的，直推式的。（1）euler 平台中 gcn_supervised 的实现是按有监督方法处理的吗？可以预测未知节点吗？（2）跟graphsage的区别只体现在邻接矩阵和聚合函数吗？（3）transductive在公式上是如何体现的？ 劳烦解答，谢谢~

Start with the gcn_supervised implementation and compare its behavior with the GraphSAGE implementation. Determine whether unknown-node prediction, the adjacency and aggregation differences, and the transductive formulation are documented or represented in the implementation; done means providing clear answers to all three questions.

Written by the indexing model from the issue text.