AIDS

该数据集名为BAMultiShapes，是对一些流行的合成基准测试的扩展，旨在评估全局解释器处理由逻辑概念组合而成的能力。它由1,000个附加了网络基序的Barabási-Albert（BA）图组成。其中，类别0包含普通的BA图以及附加了基序的BA图；类别1则包含组合了基序的BA图。任务要求基于基序的存在对BA图进行分类。

This dataset, named BAMultiShapes, is an extension of multiple popular synthetic benchmarks designed to assess the ability of global explainers to handle compositions of logical concepts. It consists of 1,000 Barabási-Albert (BA) graphs augmented with network motifs. Specifically, Class 0 includes both vanilla BA graphs and BA graphs with individual network motifs, while Class 1 contains BA graphs that integrate combined network motifs. The task mandates classifying these BA graphs according to the presence of the corresponding network motifs.