GraphSage在无监督情况下use_id的emb拼接作为feature效果会好吗？

请问NN类归纳学习，拼接id的emb的物理意义是啥？

直接学习feature的归纳表达 作为节点的emb 和 学习feature + emb(id) 的归纳表达，作为节点的emb .

区别是啥，后者会效果好些吗？

另外对于后者这种，破坏了归纳学习的特点，变成了直推学习

The issue names GraphSage and use_id but does not identify any files, tests, or entry points. Start by locating the implementation and documentation for use_id, then clarify the intended comparison between feature-only and feature-plus-ID embeddings and how inductive behavior should be evaluated.

Written by the indexing model from the issue text.