arccx

Link prediction也是分类任务，模仿节点分类任务的metric添加accuracy，f1_score，confusion_matrix，auc。其中accuracy和 F1 分数的截止阈值选择所有测试超链接分数的平均值，这一点是参考论文 [A Survey on Hyperlink Prediction](http://arxiv.org/abs/2207.02911) 中提到的： > We first considered a negative sampling strategy with α = 0.5 and β = 1 as used in NHP....

https://deephypergraph.readthedocs.io/en/latest/zh/tutorial/structure.html#zh-build-graph 这里应该是团扩展吧，我看英文文档这里是 clique expansion

**Describe the bug** dhg\structure\hypergraphs\hypergraph.py 的 add_hyperedges() 函数 ```python // 判断e_weight 类型 if e_weight is None: e_weight = [1.0] * len(e_list) elif type(e_weight) in (int, float): e_weight = [e_weight] // 这里错误...